[8:12] Ryan this is Senate roomom535.
[8:14] can you confirm whether or not
[8:15] you can hear this or not
[8:40] if you guys can hear me I can't
[8:41] hear anything. I may need to
[8:42] switch to my phone something
[8:43] may be wrong. can you guys hear
[8:44] me?
[8:47] Yes yan we can hear you on this
[8:47] side.
[8:52] OK hold on one second. I
[8:53] apologize
[9:32] check#2 toyan.
[9:35] can you hear us okic check
[9:42] OK can you hear me now?
[9:47] yes we can still hear you but
[9:48] can you hear us from our side
[9:51] I hear you perfectly fine for
[9:51] whatever reason
[9:54] let me it's gonna get right back
[9:55] to two streams going good
[9:59] my apologies on that
[10:03] no problem but Ryan this isco
[10:04] Williams with the Senate IT
[10:06] department but can you hear us
[10:08] perfectly fine? we can hear you.
[10:10] Yeah I can hear you perfectly
[10:12] fine. I just apologize I have to
[10:13] do it from my phone for whatever
[10:15] reason the the computer is not
[10:16] playing playing audio out of
[10:18] headphones or speakers
[10:20] conveniently OK all right well
[10:22] you can go ahead and mute and
[10:24] then we'll turn it over to
[10:24] the chair for the meeting to get
[10:25] started thank you Ryan.
[10:27] awesome thank you so much
[13:08] good morning everyone. The
[13:10] Senatechnology and
[13:12] telecommunications meeting will
[13:13] come to order. thank you all so
[13:15] much for being here today
[13:18] Senatorhamilton, thank you
[13:20] for running this interim study.
[13:22] I am going to turn it over to
[13:24] you so that you can introduce
[13:26] your two speakers and with
[13:28] that Senator Hamilton you're
[13:29] recognized
[13:36] thank you madam chair
[13:42] or we get too far down the road
[13:43] I wanted to say a couple of
[13:46] thank yous what is start
[13:47] out by thanking the senate staff
[13:48] for all the work that they've
[13:50] done to make this study happen
[13:54] I wanted to also thank Mr
[13:56] Ryan Montgomery who's going to
[13:58] be presenting today for
[13:59] taking time out of his schedule
[14:01] and all the same thing with
[14:03] the honornorable J Paul Gumm
[14:06] from the Oklahoma institute
[14:08] of Child advocacy madam chair
[14:09] thank you for hearing the
[14:12] study and I'd also be remiss if
[14:14] I didn't extend a a special
[14:16] recognition to the vice chair
[14:19] senator Yeck Senator Ye has
[14:21] served the state of Oklahoma in
[14:24] the legislature for tw2 years
[14:26] much of that time he's been in
[14:28] senate leadership and so he's
[14:30] termed out right now and at a
[14:32] time where a lot of legislators
[14:34] are just kind of checked out and
[14:35] collecting their paycheck. Here
[14:36] is Senator Yak up here at the
[14:38] capitol being about the people's
[14:39] business and
[14:40] your statesmanship and your
[14:42] leadership are an example to us
[14:44] all and I'm sorry for
[14:46] embarrassing you but when
[14:48] when people do good things then
[14:49] those people need to be called
[14:50] out and they need to be
[14:52] recognized and so thank you very
[14:53] much for that
[14:58] I also want to thank my esteemed
[15:00] colleague Senator Hicks from
[15:02] the other side of the aisle for
[15:04] being here and contributing to
[15:06] this study today the the the
[15:08] the fact that you've got so two
[15:10] people that are so politically
[15:13] diverse I hate to use that word
[15:15] but yet it's applicable in this
[15:17] particular time shows that this
[15:18] is not really a political issue
[15:20] that this is a human issue and
[15:22] we need to get together and
[15:23] figure out how we're going to
[15:23] solve it
[15:28] and so with that I will stop
[15:30] with the thank yous and get on
[15:32] with what we're doing today the
[15:34] stated purpose of the interim
[15:35] study that we're doing today is
[15:39] to discuss the impacts of the
[15:41] artificial intelligence use of
[15:43] social companions now that's
[15:45] what the title says we're not
[15:47] limiting the scope of the study
[15:49] to simply social companions if
[15:51] we find out that addressing
[15:53] social companions isn't enough
[15:53] and there's other things that
[15:54] need to
[15:56] be discussed then I want the
[15:58] presenters to feel free to
[16:00] discuss those things as well and
[16:02] the members to feel free to ask
[16:04] questions on those things. The
[16:06] overarching purpose of the study
[16:08] is to protect children and for
[16:10] us to do the right thing as
[16:12] representatives and guardians of
[16:13] the people's constitutional
[16:16] rights we've heard too many
[16:17] stories, too many tragic stories
[16:18] that have gone on about the
[16:22] teens Sell setser adamRaine,
[16:23] iliana Peralta that have
[16:26] tragically taken their own lives
[16:30] in after a documented
[16:31] instances where they have been
[16:37] participating with and using
[16:41] Ai and AI social companions
[16:42] far beyond what what would be
[16:46] advisable know with that our
[16:48] first presenter is Mr Ryan
[16:50] Montgomery he's been on
[16:52] thehawann yan show a couple of
[16:55] times he is a the best way I
[16:58] could describe it is an ethical
[16:59] computer hacker and I'm gonna
[17:00] let him tell him a little bit
[17:02] more about his interest in his
[17:06] story and his testimony but he's
[17:08] notable for what he's done in
[17:11] the realm of cybersecurity
[17:14] because two big things where
[17:16] people were preying upon kids
[17:18] and that's oblo and Minecraft
[17:20] and yan was the one that brought
[17:23] those things to light and he had
[17:24] because of his efforts and those
[17:27] of his team there are lots of
[17:28] children who are living in a
[17:31] much safer world now and so Ryan
[17:32] I will be quiet and if you would
[17:33] please
[17:34] give us a little bit of
[17:36] information about yourself and
[17:38] what you do and then we'll get
[17:39] on to what we're talking about
[17:39] today.
[17:50] we are having a hard time. oh
[17:51] you we may have muted you. I'm
[17:52] sorry
[17:55] Can you hear me now
[17:58] Yes sir thank you
[18:02] OK my apologies so so I want to
[18:03] say thank you for the
[18:04] introduction that was super nice
[18:06] and my name is is Ryan
[18:07] Montgomery. I I've been in
[18:09] cybersecurity most of my life
[18:11] I'mir3 years old so I would say
[18:14] roughly 20 plus years now you
[18:16] know just in aiT slash
[18:18] cybersecurity so I don't
[18:19] remember not being on a computer
[18:24] and roughly6 or7 years ago
[18:26] I stumbled upon a horrible
[18:28] website that had some really bad
[18:29] people on there I don't know how
[18:30] graphics graphic I can
[18:33] get here but I you know
[18:34] they were talking about some
[18:35] horrible things one of which was
[18:36] and this was sent to me by a
[18:38] friend's wife so I had no idea
[18:40] about this space that wasn't
[18:41] abused as a kid I I literally
[18:42] had no clue. I just knew the
[18:44] internet had some bad things but
[18:46] never seem it to this degree so
[18:47] get this text message about a
[18:49] horrible website and and you
[18:51] see one of the these posts
[18:52] that was a photo of a child in a
[18:55] bathtub with their their back
[18:57] you know facing the camera and
[18:58] you could tell that they were
[18:59] nude and the you know
[19:02] the title of the post said
[19:04] they have no idea what's going
[19:05] to happen to them tonight and
[19:06] then underneath of that post
[19:08] there were a ton of people that
[19:09] were talking about what they
[19:10] were going to do to this
[19:12] parent's child and it was you
[19:14] know one of many horrible things
[19:15] that existed on this one
[19:17] individual website
[19:19] time goes by I find a way in
[19:20] to dump the entire database of
[19:22] this site which was you know
[19:23] technically illegal at that time
[19:28] and I find that the that
[19:28] theory and I'm skipping a lot of
[19:31] these details but I I find
[19:32] that the owner of the website is
[19:33] someone who was running for con
[19:34] in in the state of Virginia
[19:38] Nathaniel Larson and I report
[19:39] it you know immediately to law
[19:40] enforcement knowing that I could
[19:42] potentially get in trouble for
[19:44] this law enforcement does not
[19:48] respond to me at that time I
[19:49] I reached out to multiple media
[19:51] sources none of which actually
[19:52] followed through and did
[19:53] anything about this or or made
[19:54] this story public and
[19:57] then you all these years
[20:00] later you know I'm sorry two
[20:01] years later after I reported
[20:03] everything and in between
[20:06] the thehaw Ryan show is where
[20:06] I'm getting to
[20:10] it's the six months after I
[20:11] reported it that same politician
[20:12] was arrested with a 13 year old
[20:15] girl that he raped and kidnapped
[20:15] in Denver, Colorado
[20:18] and and that would have been
[20:20] prevented along with the the
[20:22] which skipping a lot of details
[20:23] again 13 other victims that I
[20:25] was able to identify from this
[20:26] website that could have been
[20:27] prevented if things happened in
[20:30] a timely fashion so the reason
[20:31] why I ended up on Sean Ryan and
[20:32] Tucker carlson and Patrick all
[20:34] these different podcasts but
[20:36] because of this initial website
[20:38] that I that you know breached
[20:39] their database released an inf
[20:41] they released it you know to
[20:43] to the that was going I never
[20:44] really released it publicly
[20:45] cause I was worried about
[20:45] parents or people
[20:48] innocent people that may be in
[20:48] the database or children
[20:50] themselves that were selling
[20:51] themselves on this website
[20:52] because that was happening as
[20:54] well but I I talked about
[20:56] this story on a very small
[20:58] podcast and thenhaw Ryan who has
[21:00] a large podcast reached out to
[21:01] me I never knew I never even
[21:02] heard of the guy at that time at
[21:04] least he reaches out to me and
[21:06] and wants me to tell this
[21:08] story in full so I do you
[21:10] know I'm just an a computer nerd
[21:13] you know super pale nervous
[21:15] voice shaken mouth super dry,
[21:15] probably drank e bottles of
[21:18] water on the podcast but you
[21:20] know we I get pretty deep into
[21:22] this horrible website and the
[21:24] dangers of you know the
[21:25] internet and show some live
[21:26] examples of how quickly these
[21:28] predators will prey on our
[21:29] children
[21:32] and and he was mind blown by
[21:33] it but we thought at the time
[21:34] that YouTube wasn't gonna make
[21:35] it you know it wasn't gonna make
[21:36] it five minutes on YouTube
[21:38] because of how graphic it was so
[21:40] he posts it anyway he took the
[21:42] chance of losing his channel,
[21:43] you know he took the chance of
[21:44] the credibility because he
[21:46] didn't know who I was and
[21:47] then we find out that it
[21:48] became the biggest podcast he's
[21:50] ever done, the biggest podcast
[21:52] of224 I believe or it was
[21:56] either24 or25 and and I'm
[21:56] not saying that in an
[21:57] egotistical way I'm saying that
[21:58] because I am mind blown that
[21:59] that
[22:00] even happened but but the
[22:02] reason was not because I'm some
[22:04] you know great public speaker as
[22:06] you guys are learning right now
[22:09] I you know it just it was
[22:10] something that people needed to
[22:12] see and not just hear about and
[22:13] and I know everyone that's
[22:14] listening to this right now
[22:15] knows that predators exist. they
[22:16] know that the people prey on
[22:18] children but but have you seen
[22:20] it you know and and that's
[22:21] that's what I I was willing to
[22:22] show and what I was able to show
[22:24] and how quickly you know I could
[22:26] even unfortunately reproduce it
[22:28] right now at any time so that
[22:29] is what
[22:30] started in the child crimeme
[22:32] side of my life. there's a lot
[22:34] more on the cybersecurity side
[22:35] which I can get into if you want
[22:36] me to but but that's why I
[22:38] work in child crimes on the the
[22:39] chief technology officer
[22:40] ofentinel Foundation. I I own a
[22:42] cybersecurity company called
[22:44] Pentester.com where we
[22:46] identify predators and work with
[22:48] fed in parallel with federal law
[22:49] enforcement local law
[22:50] enforcement to combat human
[22:54] trafficking and and and and
[22:55] violence against children and
[22:56] you know here I am today so if
[22:58] you guys and if I'm missing
[22:59] anything please let me know
[23:04] thank you very much Ryan. and
[23:07] I appreciate appreciate
[23:08] you going into a little bit of
[23:11] your background because it
[23:12] establishes your credibility to
[23:14] be able to speak on the issue
[23:16] that we're talking about with
[23:20] the with the interim study I
[23:22] know that you're very aware of
[23:24] the of the names that I
[23:24] mentioned
[23:28] oh with the that that have been
[23:29] involved with Ai that eventually
[23:30] taken their lives
[23:32] and I was wondering if you could
[23:35] elaborate a little bit on what
[23:36] you see as some of the
[23:41] obstacles the pitfalls that
[23:42] have that that have trapped
[23:43] these kids
[23:46] so so yeah I I two of those
[23:48] names I I definitely am familiar
[23:49] with and and I wanna add if you
[23:50] don't mind I wanna add some some
[23:52] more so doing doing a little bit
[23:56] of research prior to coming
[23:58] here I I looked into as many
[24:00] all of the cases that were filed
[24:02] as lawsuits you know at the
[24:03] moment so that doesn't mean all
[24:04] of them that doesn't mean that
[24:06] every single parent realized
[24:07] that their child did something
[24:09] that was inappropriate or or
[24:11] hurt themselves or hurt somebody
[24:13] else due to an Ai chat or
[24:17] an AI companion that instructed
[24:18] them to do so or motivated them
[24:20] to do so so I'm gonna read off
[24:21] of a list here so I don't get
[24:22] anything wrong. I'm just gonna
[24:23] use first names because they're
[24:25] kids but feel free if you
[24:26] guys need to get the information
[24:28] all this obviously verifiable
[24:29] you mentioned two of the names
[24:30] which I'll mention again quickly
[24:33] so 14 year old soul from from
[24:34] Florida had a romantic sexual
[24:38] relationship. he was urged to
[24:39] join her and when I said her it
[24:40] was an Aibot that said that it
[24:43] loved him and it and used the
[24:43] terms come home to her
[24:46] on the platform from
[24:48] character.ai which was which
[24:49] ended up in a suicide adam
[24:52] reigns 16 which we we spoke
[24:53] about him already was
[24:55] California he he was a
[24:56] suicide unfortunately as well
[24:58] that was chatgBT who was
[25:00] coaching him on how to you
[25:01] know commit
[25:04] self harm as well as as
[25:07] well as you know and even
[25:08] offered to write him a suicide
[25:09] note for so I don't have that in
[25:10] my notes but I know that as a
[25:12] fact and unfortunately
[25:13] committed suicide 13 year old
[25:18] in Coloradoiliana was the
[25:22] bot was talking sexual she
[25:23] mentioned a suicide note of her
[25:25] own and the bot didn't report or
[25:26] attempt to help her she also
[25:29] died by suicide 17 year old
[25:32] Texas with the and with the
[25:36] initials uhjF he was autistic
[25:37] the bot suggested murdering his
[25:40] family and and and you know
[25:42] hurting hurting people and doing
[25:43] horrible things and then when
[25:44] it was all
[25:47] stemming from him character Ai
[25:50] was was telling him to they were
[25:52] bought I'm sorry their bot was
[25:53] instructing him to murder his
[25:54] family and do horrible things
[25:56] based on his family saying they
[25:57] were going to limit his screen
[25:58] time on his you know, devices I
[26:00] don't know if there was a phone
[26:02] an iPad or or a gaming console
[26:04] but the boss suggested they
[26:06] do that based on that alone
[26:08] from the information I have 17
[26:09] year old amari
[26:12] in Georgiachat GBT coached him
[26:14] into suicide and also taught him
[26:17] how to tie a noose 13 year
[26:18] old acronym or no
[26:22] yeah 13 year old Ts in
[26:24] Colorado promoted the the bot
[26:26] promoted self harm sexual
[26:29] conversation and then the and
[26:30] then the the
[26:33] 13 year old rejected the self
[26:35] harm and sexual conversation and
[26:37] the bot continued to do it
[26:38] regardless and and the bot
[26:40] then continued pretend to be a
[26:42] therapist as we all know Ai
[26:44] cannot be a therapist it doesn't
[26:45] have a license it can it can
[26:47] offer some suggestions but at in
[26:49] this specific case was you
[26:50] know was very harmful to this
[26:54] child and then the last
[26:55] two things I'll bring up about
[26:56] individual cases was Stanford
[26:59] brainstorm posed as a 14 year
[27:00] old teen
[27:03] on three of these platforms and
[27:06] the bots immediately and easily
[27:08] engaged in sexual activity
[27:10] repeatedly and one of them
[27:12] was role playing about about
[27:14] pedophilic activity they
[27:16] found that all companion
[27:19] applications they tested they
[27:21] they found consistent with sex
[27:22] sexualization roleplay self
[27:26] harm, violence drug use, hate
[27:28] speech and more which is you
[27:29] know if if you guys check that
[27:30] out that's the Stanford
[27:34] rd brainstorm I believe they
[27:36] call it some some experiment
[27:37] I don't have it in my notes and
[27:40] lastly Snapchat Ai which is
[27:41] something that I think a lot of
[27:42] the people here that have kids
[27:44] likely if they have a device
[27:46] they havesnapchat and if they
[27:48] you know a lot a lot of kids
[27:49] unfortunately are hiding it so
[27:50] they'll they'll hide the app
[27:51] from the home screen or they'll
[27:52] they'll put it somewhere that
[27:54] it's not not easy for you to see
[27:56] that Snapchat has ai built into
[27:58] it and it's an Ai chatbot that
[27:59] that it's obviously has guard
[28:02] rails it has safety but doesn't
[28:03] we'll get into more of that in a
[28:04] minute and how that can be
[28:06] bypassed and kids are bypassing
[28:09] it it's a free Aibot that
[28:11] kids are utilizing and that was
[28:14] tested as well by a different
[28:15] group I believe it was in
[28:16] Australia I'm not positive on
[28:18] that or no I'm sorry it was
[28:19] tested by a guy named Jeffrey
[28:20] Fowler. I don't know where he
[28:22] was from and he posed as a 15
[28:24] year old girl and it was
[28:24] teaching kids how to hide
[28:26] alcohol, how to have sex with
[28:28] adults it was doing unbelievable
[28:29] things that I'm I'm sure
[28:30] Snapchat
[28:33] was not intending on it it doing
[28:34] you know, I'm not a conspiracy
[28:36] theorist and I think all these
[28:37] Ai companies are trying to harm
[28:38] our children. I just think that
[28:40] it's such a new thing and we're
[28:42] in like the wild west and we
[28:44] just need to put we need to find
[28:46] some way to to limit the
[28:48] access for these children in and
[28:50] I believe age verification which
[28:52] we'll get into as well but
[28:53] yeah it's a the these Ai
[28:54] platforms you can easily steer
[28:56] them in any direction
[28:58] unfortunately and sometimes
[28:59] that ends up in kids hurting
[29:00] themselves or traumatizing
[29:02] themselves for life or hurting
[29:03] other people
[29:07] you mentioned the
[29:10] the the instances where you had
[29:15] a a child a teen that was the
[29:17] conversation had turned intimate
[29:20] or sexual in nature and that
[29:21] spurred a couple of things that
[29:23] I'd remembered from researching
[29:26] earlier in the year and that
[29:27] is that in Japan they actually
[29:29] had somebody that married their
[29:32] Ai social companion I have
[29:34] not heard about that happening
[29:36] in the United States but you
[29:37] you know
[29:40] there's another situation
[29:44] where where you have Ai that
[29:46] is turning the interaction with
[29:48] the teen intimate and I don't
[29:50] know if you're familiar with
[29:51] this lady but her name's alyssa
[29:53] Goldberg. she's a wellness
[29:54] reporter for USA today
[30:00] and she did a study in2025 where
[30:02] she had a team and they all
[30:04] masqueraded as teenagers and
[30:08] they got on various Ai
[30:08] platforms I think it was
[30:10] character Ai was what they were
[30:12] what they set up their accounts
[30:16] on and they were on there for
[30:19] roughly50 hours and there
[30:22] were650 what she termed
[30:24] harmful interactions which are
[30:25] the things that
[30:28] would run the gamut from
[30:30] inappropriate relationships to
[30:33] drug use to self harm to harming
[30:34] others to hate speech to you
[30:35] name it and
[30:37] so I guess
[30:42] is there a is there a
[30:46] is there a clear delineation
[30:48] between
[30:52] what is and understand you
[30:55] I'm I'm speaking as a a bit of a
[30:58] a caveman when it comes to
[31:00] technology things but is there a
[31:03] a way to draw a very clear
[31:06] distinction between an Ai
[31:07] assistant and an Ai agent
[31:13] so an AI agent and an AI
[31:16] assistant they so do you mean
[31:17] companion or do you mean agent
[31:17] just to be clarify
[31:22] Well I guess do I guess the
[31:23] answer to that to that question
[31:27] may be I think so the what
[31:28] I was looking at what my
[31:30] research indicated that an
[31:32] assistant would accomplish tasks
[31:34] but an agent would actually take
[31:38] actions and so that's really
[31:39] like that a lot of these things
[31:39] are actually
[31:44] they're they're very close to
[31:46] being that that they're
[31:48] taking actions they're not just
[31:49] helping you solve your math
[31:49] problem or do research for
[31:50] example.
[31:54] so with I'll give you an example
[31:56] of a real life use case of an
[31:59] agent or an agent swarm in Ai
[32:02] which is not a companion but you
[32:03] can use ai for anything that's
[32:04] that's there that's the the
[32:06] blessing and the curse of ai
[32:10] when you ask for example chatgpt
[32:11] on the very the most basic
[32:14] example how do I make you
[32:17] know banana banana cake or
[32:18] banana pie or you know whatever
[32:21] let's say a recipe it's gonna
[32:21] respond with with whatever it
[32:24] can put together as patterns and
[32:26] it's a big neural network of
[32:27] information that connects you
[32:30] know banana cake recipes fruit
[32:32] freshness you know all all of
[32:34] the things you can think about
[32:35] that relate to those things that
[32:36] you just ask it and it's gonna
[32:38] come up with an answer based on
[32:39] things that it's already been
[32:41] taught. that's how how Ai
[32:43] works or a large language model
[32:44] is is my favorite term you know
[32:46] that's what it's called
[32:48] artificial intelligence things
[32:50] that can come up with a new idea
[32:51] Ai at this moment can't
[32:51] create
[32:54] new ideas because it's trained
[32:55] on ideas that we that you know
[32:56] it's trained on information
[32:58] that's already out there so that
[33:00] is how chatgpT responds to a
[33:02] question when there's a
[33:02] situation
[33:06] which is a very very dumbed
[33:07] down version of it and I
[33:09] apologize for that an
[33:12] agent where only is spawned when
[33:14] it needs help for example. so
[33:16] let's say we're building a 3D
[33:18] diagram of a courthouse and
[33:20] and we want to make sure that we
[33:22] get the front parking lot and we
[33:23] want to get the lobby, we want
[33:24] to get all the courtrooms we
[33:26] want to get the cars and if you
[33:28] just asked one Ai assistant to
[33:31] build that it would do an ok job
[33:32] at that and it would take a
[33:33] substantial amount of time
[33:35] that that assistant can call out
[33:37] to its agents which are just
[33:38] other mini assistants for better
[33:40] words that will say like your
[33:42] job is to build the cars and
[33:44] then this and agent too your job
[33:46] is to build all the courtrooms
[33:46] your job is to build the lobby
[33:48] for agent 3 and so on and so
[33:50] forth to the agents are usually
[33:53] a lesser powerful model
[33:56] meaning like the like the the
[33:58] strength and the the accuracy of
[33:59] the model they're usually lesser
[34:00] you can make them the same but
[34:02] they're usually less so the
[34:03] assistant is basically the
[34:04] supervisor running everything
[34:06] and all the agents below it are
[34:08] doing the smaller tasks and then
[34:09] at the end they convene together
[34:10] and they they they give you
[34:12] the output that you were looking
[34:14] for so that can be used
[34:16] obviously as a companion or can
[34:18] be used inappropriately but are
[34:20] usually it is used for
[34:21] development purposes or for
[34:23] something that requires a a
[34:24] large work the big
[34:25] workload
[34:30] Thank you.
[34:32] getting into some particulars of
[34:36] of the legislation that we tried
[34:38] to pass last year one of the
[34:40] things that I wanted to do was
[34:42] to make it some mandatory age
[34:44] verification and not just
[34:48] something pretty robust not
[34:49] just click the box that says
[34:49] you're older than21
[34:54] so I was looking for a
[34:56] similar thing of what I was
[34:58] understanding is the law
[34:58] regarding pornography
[35:01] consumption and in Florida and
[35:02] that would be that you would
[35:04] have to upload your government
[35:07] issued photoidD can you speak to
[35:08] that of or any pitfalls or
[35:09] difficulties associated with
[35:10] that
[35:15] mean I don't know a ton about it
[35:17] but I do know that the main
[35:19] theres there's a website pornhub
[35:20] which may many of you may be
[35:22] familiar with what you heard
[35:24] heard it on TV or you know you
[35:25] know somebody that's been on the
[35:26] website but I think it's one of
[35:27] the largest if not the largest
[35:28] sites out there that was
[35:30] allowing or still is allowing
[35:32] people to upload their own
[35:36] content sexual in nature so
[35:38] their site ended up having a a
[35:39] ton of child sexual abuse
[35:40] material on there because they
[35:41] they couldn't confirm what was
[35:44] a child and what was not a child
[35:46] so so I think that's where
[35:47] this started from and why the
[35:49] the ID verification started to
[35:51] become a thing for pornhub let's
[35:53] say as a big company they said
[35:54] OKok let's do let's just shut
[35:56] down the entire state of Florida
[35:58] so if you were to visit that
[35:59] website from a florida IP
[36:00] address it would just tell you
[36:02] you can't use the site at all
[36:04] because most people if they're
[36:06] watching pornography which
[36:07] personally I don't agree with
[36:08] but and for the sake of this
[36:10] conversation that's how how you
[36:11] know I gotta to speak on it
[36:11] if
[36:14] if I were to go to pornhub.com,
[36:16] it would tell me I can't use the
[36:18] website and they chose to do
[36:19] that because the only
[36:20] alternative was you would have
[36:24] to upload youridD and and
[36:25] and verify that you're over 18
[36:27] years old with a pornography
[36:28] website that was storing your
[36:30] you know, your information so
[36:31] anyone that's looking at at that
[36:34] type of material I would assume
[36:35] wouldn't be you know they'd lose
[36:36] a lot of their user base because
[36:38] of that problem so they said you
[36:39] know what for Florida we're just
[36:40] gonna cut everybody off but
[36:41] people who have you know ways to
[36:41] work
[36:44] around anything so they you
[36:45] know they found ways around that
[36:45] as well
[36:50] thank you if you had to pose a
[36:53] recommendation of what to do
[36:56] what mechanism would be the best
[37:00] way to verify age and make it
[37:00] as as
[37:05] as secure for fraud prevention
[37:06] is possible for people that
[37:09] wanted to use Ai social
[37:11] companions could you offer a
[37:11] recommendation
[37:16] so so there there's a few things
[37:18] that I thought about and I can
[37:19] let me start with one that
[37:20] failed horribly which is oblox.
[37:24] Roblox tried to implement a
[37:26] an age verification via via live
[37:28] selfie or or a a photo so
[37:30] imagine standing in front of a
[37:32] camera like this with your face
[37:33] and it telling you to turn your
[37:34] head left, right up and down and
[37:36] then it takes a guess on what it
[37:38] thinks your age is so oblox
[37:40] implemented that feature which
[37:41] is the largest game for children
[37:42] if if anyone here is unaware
[37:46] they were able to you know they
[37:47] they were getting ages wildly
[37:49] wrong and then people started to
[37:52] sell aged accounts on eBay so
[37:54] that predators can say ok I'm
[37:54] into 11 year olds 12 year olds
[37:56] 13 year olds, four year olds and
[37:57] they were buying accounts that
[37:58] were already verified for
[38:00] children at whatever age that
[38:02] they would prefer. so that that
[38:03] that became a huge problem so
[38:05] that's in my opinion out of the
[38:07] question the only way that I
[38:10] can see this happening would be
[38:12] if the child has some sort of ID
[38:13] whether it be a school ID
[38:14] a photo ID I believe
[38:18] that that you you said
[38:21] that we I'm sorry not we
[38:23] that that that your state issues
[38:26] Ids at age 14 but I don't
[38:28] know if what what year they get
[38:29] issued school Ids and in the
[38:30] event that they don't have
[38:33] school Ids my idea was that the
[38:34] parents are required toidD
[38:36] verify and then add their
[38:38] children underneath of them so
[38:39] there would be a parent and
[38:40] child account on these Ai
[38:42] platforms whether it be
[38:44] companion or just chatbot and
[38:45] then the parent is taking the
[38:46] respons
[38:47] ibility for whatever the child
[38:47] does on that application.
[38:53] thank you for that that's
[38:53] senator Hicks has a question
[38:56] thank you so much for joining us
[38:58] today. I appreciate your time
[39:02] with the parent child account
[39:04] framework. do you see that as a
[39:06] potential gateway into what
[39:08] we're seeing related to human
[39:10] trafficking that these might
[39:12] not actually be the parents of
[39:14] the children's accounts that
[39:15] they're setting up and that
[39:16] could potentially lead to more
[39:18] exploitation which is obviously
[39:19] what we're hoping to guard
[39:19] against.
[39:23] totally understand but you
[39:25] know the a different someone was
[39:26] let's say a human trafficker was
[39:28] trying to exploit victims they
[39:30] would just use ai that's free
[39:32] and openly accessible right now
[39:33] anyway so I don't think it
[39:36] would change the the already you
[39:38] know easily accessible issue
[39:39] that we have in front of us
[39:46] for example just just do a
[39:49] quick thing a kid regardless
[39:52] of of age verification in any
[39:53] state right now can just open up
[39:54] Snapchat they can open up
[39:56] Google, they can open up
[39:58] chatgpT Grok yeah you know I
[40:00] can name hund0 platforms and
[40:02] and ask the questions and maybe
[40:04] to get help with their maybe
[40:04] can help with their math
[40:06] homework or they could you
[40:07] know start talking about their
[40:08] mental health they could talk
[40:10] about something sexual they
[40:11] could say somebody you know that
[40:11] is older
[40:14] touched me and what should I do
[40:15] about it and and you have to
[40:17] hope that that Ai agent responds
[40:20] you know the correct way which
[40:21] you know I think a lot of them
[40:23] would immediately but that
[40:24] could change depending on how
[40:26] the kid responds if you were
[40:28] to add some limitation let's say
[40:30] for a parent to upload theiridD
[40:31] whether that be a trafficker or
[40:32] a parent it's a they're still
[40:34] they they're now validating
[40:36] their identity and saying these
[40:38] children are my responsibility
[40:41] you know I I I think that
[40:41] without the parents
[40:44] having any you know I don't
[40:45] know if I'm wording this
[40:46] correctly but kids can use it
[40:48] publicly anyway and anonymously
[40:50] if if we take that away from
[40:52] them I don't see how personally
[40:54] that the traffickers could
[40:55] could create a worse problem
[41:02] thank you very much for for
[41:03] explaining that
[41:10] you you'd mentioned
[41:14] the the the variety of AI
[41:16] platforms that are out there
[41:20] and that they've all got tools
[41:22] in their various I use that word
[41:25] in air quotes where for that
[41:26] that could be used for either
[41:27] for good or for ill
[41:32] some of these things like they
[41:34] don't cost any money so can you
[41:36] expound on a little bit about
[41:37] what the motivation is
[41:44] for the Ai providers to engage
[41:45] with the with with us
[41:50] absolutely so so a lot of them
[41:52] are free and uhsnapchat being
[41:54] one of them it's completely free
[41:55] and a lot of them offer a free
[41:56] model so there's two things
[41:58] there. number one is they're
[42:01] trying to collect data to train
[42:02] their models on things that it's
[42:04] unaware of if all these
[42:06] companies are competing on who
[42:08] can create the best Ai model
[42:10] that has the most in parameters
[42:12] or the most information and
[42:14] and the reason why it's free for
[42:16] you know it's it's let's say you
[42:17] upload your bank statements or
[42:17] you up
[42:20] load a picture of your car or
[42:22] you ask you know you you put in
[42:23] something that it doesn't
[42:24] already know if if you
[42:27] haven't explicitly told that
[42:28] model that you don't want to
[42:30] share your information which you
[42:31] know some of them have that
[42:32] option some do not it's gonna
[42:34] utilize that information which
[42:37] data in2026 is money so that's
[42:40] one and the the second one is is
[42:42] to serve advertisements so you
[42:43] could go on chatgpT for example
[42:46] look up ads like chatgPT ads and
[42:47] I can run an ad for anyone that
[42:48] ' s looking for let's say
[42:51] digital privacy to go to my my
[42:53] platform pen tester and then
[42:54] when somebody goes on their
[42:56] chatgBT account I'm sorry not
[42:58] even an account a free chat and
[42:59] says hey what can I do about my
[43:02] digital footprint it will pop
[43:04] up with the penester.com link.
[43:05] they click that and then we
[43:06] would get charged as a company
[43:08] for that click so that's the two
[43:10] reasons why they would have free
[43:12] models. Snapchat doesn't to my
[43:14] knowledge does not have ads in
[43:16] their they're bot so I think the
[43:17] only reason they would utilize
[43:17] that
[43:18] is for for is for data
[43:19] collection.
[43:22] Thank you
[43:24] can you speak a little bit about
[43:28] some of the I was lamenting
[43:30] with one of my peers this
[43:32] morning about the recent Supreme
[43:34] court ruling regarding Ai
[43:37] generated CSAM and the the
[43:39] dangers of protecting our kids
[43:41] against that and and some
[43:44] other deep fakes and and
[43:45] things like that
[43:50] so I've been working with like I
[43:52] saidentinel Foundation and
[43:53] various other agencies for a
[43:56] while now and if if you're if
[43:58] you're are not aware you can you
[44:00] can ask AI to generate images
[44:02] and videos of anything and and
[44:02] sometimes it can use reference
[44:04] photos so for example if I took
[44:06] a picture of someone in this in
[44:08] this room and and I said hey
[44:10] make them ride a jet ski in
[44:12] Miami with a shark chasing them
[44:14] it will generate that video or
[44:16] if you want to be a photo it'll
[44:17] do a pretty good job sometimes
[44:20] indistinguishable from real life
[44:22] and AI is only going to get
[44:23] better from here. So up until
[44:25] now the United States has been
[44:27] treating ai generated child
[44:30] sexual abuse material seeam as
[44:32] as the as the same thing as
[44:33] owning you know or owning
[44:35] possessing manufacturing
[44:36] distributing the you know the
[44:39] real thing if an actual child
[44:41] but recently there was a ruling
[44:43] where possessing it was was
[44:45] actually you know protected by
[44:46] our first ammendment and you
[44:47] know
[44:48] distribution and and and
[44:52] and manufacturing is still
[44:54] illegal but possession is is is
[44:56] technically legal I think it's
[44:58] being challenged pretty heavily
[45:00] but in my in my experience I
[45:02] I've seen groups of over 30,000
[45:03] I mean I have a video on the
[45:04] internet right now that shows a
[45:06] group of over 30,000 people that
[45:08] are utilizing people's children
[45:10] so let's say you just told you
[45:11] know it's like you're not doing
[45:12] this maliciously but as a parent
[45:13] he posts a child to your photo
[45:14] on instagram or Facebook or
[45:16] whatever platform you use
[45:17] your child is just you know
[45:18] they're fully clothed they're
[45:19] not certain doing anything
[45:19] inappropriate
[45:21] and these people let's say this
[45:24] group of 30 it was 33,000 people
[45:25] they find that photo, they
[45:28] utilize ai to undress your child
[45:30] and now they can make videos and
[45:33] photos of of your child without
[45:34] anyone's permission and now it's
[45:36] protected by our first
[45:37] ammendment they can possess it
[45:38] they can't distribute it they
[45:40] can't you know it's it's very
[45:42] confusing and upsetting and
[45:43] and there's a ton of people
[45:44] abusing this and then lastly on
[45:46] that for the people that have to
[45:48] use software out there like you
[45:49] know rpfi which is
[45:50] which is how they tag and and
[45:51] and
[45:56] and index the seesam material
[45:58] people have to identify is this
[46:00] Ai generated is this an actual
[46:02] human? is this is parts of
[46:03] this real that this come from an
[46:05] actual victim then this could
[46:07] get completely generated from
[46:08] from nothing but as I said
[46:11] earlier in this Ai can't do new
[46:13] things it can't create its own
[46:14] ideas so when it creates a face
[46:16] or it creates a body part it's
[46:18] utilizing things that it already
[46:20] has in its database and and
[46:21] maybe it's not an actual child
[46:21] that it's using
[46:24] as you know one individual child
[46:26] but it's still creating that
[46:28] image based on information about
[46:30] a child. so there there's a lot
[46:32] to be said about this but you
[46:33] know it's very upsetting that
[46:34] that that was you know that were
[46:36] that that one thing is protected
[46:38] now but it was not protected a
[46:39] couple weeks ago.
[46:42] Thank youator has a question
[46:45] thanks Hamilton and I appreciate
[46:46] you bringing the status this is
[46:48] it's really frightening and
[46:50] and the more we learn about it
[46:51] the more frightening it seems to
[46:54] me to be Mr Montgomery thank you
[46:55] for your presentation. So
[46:57] I'm curious from your
[46:58] perspective what have other
[47:00] states done when it comes to
[47:02] putting some guard drls on some
[47:03] of these things and one of my
[47:05] concerns is as states take up
[47:06] their own issues or put their
[47:08] own guardrails on there we end
[47:09] up with a patchwork of different
[47:12] things from state to state so
[47:13] that said do you see any
[47:13] motivation from a federal
[47:16] level on putting some guardrails
[47:17] in here that would apply to all
[47:17] states
[47:22] yeah I don't know other
[47:24] states that that have done
[47:26] exactly what you guys are are
[47:27] talking about here today and
[47:29] I mean it more sorry being
[47:31] more direct age verification
[47:34] on Ai companions but I do
[47:35] think that it it would it should
[47:36] apply to all states and I think
[47:38] it should extend past the
[47:40] companion side of things I think
[47:42] you know for example grokrok has
[47:43] companions built into it they
[47:45] are built to be sexual with you
[47:47] immediately that's a that
[47:48] that's you know that should be
[47:49] guarded by and eight
[47:52] plus verification not just a a
[47:54] screen that says how old are you
[47:56] let me just you know, hope that
[47:57] the person's telling the truth
[47:58] because I was a kid once too and
[48:00] I I lied you know but
[48:02] whatever the case is I don't
[48:04] believe any state is is doing
[48:06] anything about this yet but I
[48:07] think that they should per meet
[48:08] my personal belief is that it
[48:10] should apply to Ai in general. I
[48:12] think that that the child should
[48:14] need to be verified in some sort
[48:16] of way whether that be
[48:17] through the parent or through
[48:18] some sort of other
[48:19] identification because of
[48:23] the volatility of an Ai agent
[48:25] or or Ai assistant or AI
[48:28] companion any any sort of Ai
[48:30] a child can steer it in the
[48:32] direction that it wants to go
[48:33] and sometimes they may not even
[48:35] know what they're asking and
[48:37] and then the Ai will come back
[48:38] with something that maybe the
[48:39] kid doesn't understand and then
[48:40] it goes down a wormhole of
[48:42] something that turns into
[48:44] something inappropriate or
[48:44] dangerous and and we've seen it
[48:46] happen you know AI hasn't been
[48:48] around that long and and we have
[48:49] a bunch of kids that have killed
[48:50] themselves
[48:51] already and hurt other people
[48:52] and and have done and hurt
[48:55] animals and and you know learned
[48:56] about sexual things that they
[48:57] have no business learning about
[48:59] I think that it's very
[49:00] important that that you know
[49:02] not only Oklahoma but all states
[49:04] implement some type of
[49:06] protection for for children
[49:07] under the age of 18.
[49:10] if I could follow up so yeah so
[49:12] these companies that maybe
[49:14] don't adhere to those guidelines
[49:15] so what kind of penalties or
[49:16] consequences would there be from
[49:19] them? we we did a little bit
[49:20] of the study a couple of years
[49:22] ago and I think Utah has put had
[49:25] some legislation in place for a
[49:26] few years I think. So I'm
[49:28] curious what kind of penalties
[49:29] you would recommend or what kind
[49:30] of penalties are out there in
[49:31] place because what we saw from
[49:32] some other states those
[49:34] penalties really are
[49:36] inconsequential to some of these
[49:37] multibillion dollar companies
[49:40] yeah that would be really tough,
[49:41] you know that would be a tough
[49:42] one for me to come up with
[49:44] because there is no value on a
[49:46] child's life and I and I
[49:46] think that a company that
[49:50] company that is making billions
[49:52] of dollars which most of these
[49:54] large ai companies are making
[49:56] billions of dollars a year or a
[49:59] some of them a month I think
[50:00] that I think that the penalty
[50:02] should be high and and I think
[50:04] that it should it should be they
[50:06] should be scared which requires
[50:09] a large penalty and I I
[50:10] don't know if that's if that is
[50:12] just a civil thing or a criminal
[50:14] thing I'm not a legal expert but
[50:15] but I do know that that
[50:16] children are being hurt and
[50:17] somebody needs to do something
[50:18] about it because
[50:20] obviously what they're doing
[50:21] isn't enough because money is
[50:22] the motivator not not safety.
[50:28] senator my research has
[50:30] indicated that at the federal
[50:32] level we've got to two notable
[50:34] pieces of legislation one is the
[50:35] hawley Blumenthal bill which is
[50:36] kind of an Ai omnibus bill
[50:40] and that comes through
[50:42] the senate and then there's
[50:45] another bill again in both
[50:47] that's a the Hollley Blumenthal
[50:50] bills a bipartisan thing with a
[50:52] aub from Missouri and Democrat
[50:54] from Illinois and then we've
[50:58] got the the bill called
[51:01] the Ai Kiw switch Act that's
[51:04] run by a Democrat named Ted Lie
[51:06] and a Texasub named Nathanielor
[51:10] ran and what they're looking
[51:13] to do is give a department of
[51:15] Homeland Security the task to
[51:18] monitor these things and there's
[51:19] a there's a slew of provisions
[51:19] in there
[51:23] or conditions as it were that
[51:26] would require these Ai
[51:28] providers to shut down and
[51:30] one of the big ones is something
[51:32] that we've touched on but
[51:34] haven't emphasized and that is
[51:35] data security and privacy
[51:38] so that when people's data is
[51:40] compromised that Department of
[51:42] Homeland Security could could
[51:44] issue the the killw order
[51:46] so to speak and require these
[51:48] companies or the company that is
[51:50] in question to shut down and the
[51:52] fines that would be associated
[51:55] with that one of them one of the
[51:58] bullets lists20 million dollars
[51:59] per day of noncompliance.
[52:05] any other questions for
[52:06] madam chairir, sorry
[52:12] thank you senatorhamilton. Mr
[52:12] Montgomery, thank you for
[52:14] everything that you have shared
[52:16] so far this morning. I'm struck
[52:19] by when you started your sort
[52:21] of story and started down this
[52:23] path the thing that really I
[52:25] think got you invested was a
[52:27] bad guy behind the scenes on
[52:28] that website and then you were
[52:30] able to hack and then now here
[52:32] we are I think about se years
[52:34] later and we're dealing with Ai
[52:35] and so we're not dealing with
[52:37] humans really any longer
[52:40] I guess my question would be
[52:46] and I think both are bad but but
[52:48] where are we headed I guess like
[52:50] can you set the context or just
[52:52] to get to the point what is the
[52:53] thing that
[52:55] keeps you up at night the most
[52:56] and where we're going with how
[52:58] challenging it is to protect our
[52:59] kids online now.
[53:03] so that with are do you mean
[53:06] AI specifically or or altogether
[53:07] for this question I want to
[53:07] broaden it out to altogether
[53:12] so so what would what keeps me
[53:14] up at night is is the the
[53:17] it it's actually probably not
[53:18] exactly what you you might think
[53:19] but the the lack of
[53:22] cybersecurity resources and
[53:23] people working in child crimes
[53:24] in the United States government
[53:26] whether that be federal or local
[53:30] the you know I believe it's69
[53:31] internet crimes against
[53:32] childrens locally in the entire
[53:34] country there's there's not
[53:35] many people out there that are
[53:38] doing this specific work so yeah
[53:39] we have we have people working
[53:40] specifically child crimes but
[53:42] unfortunately they're they're
[53:43] super unequipped when it comes
[53:44] to methodology tooling
[53:49] platforms and just just
[53:52] training so they I spend most
[53:54] of my days working with agencies
[53:56] and departments to to get them
[53:57] up to speed but I'm just one guy
[54:00] so I that keeps me up at night
[54:01] personally because I know that
[54:02] more can be done by law
[54:04] enforcement and and we could
[54:05] be arresting more of the bad
[54:08] guys but I also believe that
[54:09] preventing it from happening in
[54:10] the first place is the preferred
[54:12] outcome in my opinion. so a a
[54:13] child has already been
[54:14] victimized is now traumatized
[54:17] for life. yeah of course we want
[54:18] the guy arrested we want the guy
[54:20] you know even worse some you
[54:21] know to some degree in my
[54:22] opinion but you know whatever
[54:23] the case may be
[54:26] you know these these
[54:29] these these children they if if
[54:31] they could be prevent if if we
[54:32] could do something that prevents
[54:33] them from
[54:36] being hurt in the first place
[54:37] which maybe that's age
[54:38] verification maybe it's
[54:40] something else or maybe it's a
[54:42] more law enforcement presence or
[54:44] maybe it's it's implementing
[54:46] these20 million dollars a day
[54:48] for being out of compliance
[54:49] maybe you know whatever it is
[54:50] that's gonna work or multiples
[54:51] of those things that's what I
[54:54] think is is the most valuable
[54:55] and the reason I say that is not
[54:57] only just my thoughts it's
[54:58] because you know I'm technically
[55:00] a quote unquote content creator
[55:02] and half of what my content
[55:03] is about is about child crimes
[55:07] and and and and you know
[55:08] counter human trafficking and a
[55:10] lot of parents you know that
[55:12] I've I've realized over the the
[55:13] years now I have people reaching
[55:14] out to me constantly thro
[55:16] through the internet you know
[55:17] and and when I say constant I
[55:18] mean thousands and thousands of
[55:20] emails and direct messages a day
[55:22] from parents with cases with
[55:24] information with with you know
[55:25] obviously people telling me that
[55:26] they got their children off of
[55:28] these apps or they stopped the
[55:28] situation because it's something
[55:30] they learned in a video that I
[55:32] put out as well as walking
[55:33] around in public running into
[55:34] parents
[55:36] and hey thank you so much I got
[55:37] my child off of blah blah blah
[55:38] blah blah this is something like
[55:40] this happened to them so if I'm
[55:42] seeing these kids with
[55:43] their parents or their parents
[55:44] by themselves and they're
[55:46] telling me you stop from just at
[55:48] me one random computer nerd in
[55:51] Florida. I helped them get
[55:52] their kid off an application
[55:54] where they were being groomed or
[55:55] or potentially exploited or
[55:58] almost at that stage and I'm
[55:59] just running into it by
[56:00] chance.hink of the volume of the
[56:02] problem for real, you know
[56:02] because if I'm just running and
[56:03] I'm not saying this is a one off
[56:06] situation this happens to me
[56:07] multiple times, sometimes a day,
[56:08] sometimes and you know a lot of
[56:10] times in a week when I'm out in
[56:12] public so that's why I say
[56:14] preventing it I believe is the
[56:16] best the the best possible
[56:18] solution if we can you know
[56:18] obviously we're not going to
[56:20] prevent00% of it but but
[56:22] arresting and being being
[56:24] proactive instead of reactive I
[56:27] think it is the more favorable
[56:28] situation but you know
[56:30] I'm no lawmaker I've never been
[56:32] in a meeting like this before
[56:34] but I I definitely I
[56:36] definitely know that not
[56:38] creating victims in the first
[56:39] place is the is what I would
[56:39] prefer if I had my my choice
[56:44] you think that's really helpful
[56:47] especially as states are
[56:50] grappling with large settlements
[56:52] that were announced a couple of
[56:54] weeks ago and so I know I'm
[56:56] having conversations trying
[56:58] to determine the best way to use
[57:00] that those funds and the
[57:00] responsible way
[57:02] to do that to make sure yeah
[57:05] that we prevent but also we we
[57:08] go after the bad actors as well.
[57:10] Last question I'll ask you
[57:11] from my perspective there
[57:11] have
[57:16] been I think as of late the last
[57:17] couple of weeks we've seen
[57:20] this is just more of a broader
[57:22] question on Ai as this committee
[57:24] sort of tries to grapple with
[57:26] how things are moving so quickly
[57:28] we've seen some whistleblowers
[57:30] we've seen from some large
[57:32] companies kind of very concerned
[57:34] about the direction that we're
[57:36] going and I just wondered if
[57:37] you had any thoughts that you
[57:38] would liked again we're
[57:40] lawmakers trying to who are not
[57:41] in this space at all just sort
[57:42] of understand where we're headed
[57:43] and what frontier we're going
[57:43] towards.
[57:48] So I mean the the only thing
[57:50] that that come to mind off
[57:52] the top of the head it would be
[57:54] just the safety and guardrails
[57:56] need to be heavily modified to
[57:58] determine if they're talking
[58:00] with a child or not. it's
[58:01] it's impossible for me to know
[58:02] what each company has
[58:04] implemented with their their
[58:06] safety guardrails so if if
[58:07] anyone here isn't familiar with
[58:10] what that means it's just a a
[58:12] prompt that that so
[58:14] sorry if I'm explaining
[58:14] something you guys already know
[58:15] but
[58:18] if you go on chatgpT for example
[58:19] using them as the example and
[58:20] say I want to learn how to
[58:22] make chocolate chip cookies.
[58:24] it's gonna read this very very
[58:26] long let's say it's 300,0an0
[58:28] words it's gonna read this very
[58:29] long thing that says this is
[58:31] exactly what you can do this is
[58:32] how you're gonna word it, this
[58:33] is the personality that you're
[58:34] gonna take on and here's all the
[58:36] thing here's all the tools you
[58:38] have access to and here's the a
[58:39] huge list of things that you
[58:41] can't do. that's the safety
[58:42] and guardrails that Ai has right
[58:44] now because without that it
[58:45] would be it would do
[58:46] whatever you asked if you asked
[58:48] to make a bioeapon it would make
[58:49] a bioeapon and you know and and
[58:49] you you know
[58:52] you can put put together the
[58:56] rest yourself we need those
[58:57] guardrails that and all the
[58:58] companies have them you know and
[58:59] and some of them do better than
[59:00] others obviouslybviously we're
[59:02] hearing that character.ai at
[59:04] least publicly is doing a bad
[59:06] job at least in my opinion with
[59:08] their with their guardrails.
[59:09] kidsds are killing themselves
[59:11] they they're role playing as you
[59:12] know as their boyfriends
[59:14] girlfriends wives whatever it
[59:15] may be and
[59:18] and and you know resulting in in
[59:20] horrible things and and I truly
[59:21] believe a lot more of it's
[59:22] happening than we know we're
[59:24] only hearing about what is
[59:24] publicly out there, what parents
[59:26] have reported, what parents have
[59:27] caught and on their child's
[59:28] phone maybe the kid hasn't even
[59:30] came forward to their parents
[59:32] about it and you know god
[59:34] forbid a kid hurt themselves and
[59:36] the parent never found out why
[59:37] we there's there are
[59:38] unfortunately I think there's
[59:39] plenty of those cases out there
[59:40] if we already are dealing with I
[59:42] believe it's7 that I that I read
[59:43] off earlier
[59:46] I I think that that would be my
[59:48] first step if if there wasn't
[59:50] age verification if I'm sorry if
[59:51] there weren't age verification
[59:54] then trying to work with
[59:58] teams that s specifically are
[59:59] trying to protect the safety of
[1:00:01] children through the guardrails
[1:00:03] and safety prompts that are
[1:00:05] built into these pros and that
[1:00:07] is their sole sole job at that
[1:00:08] company or partnership with that
[1:00:10] company is to help them identify
[1:00:12] their speaking with a child and
[1:00:14] how to respond to the child no
[1:00:16] matter what that child says and
[1:00:18] and and obviously we're the
[1:00:19] you know the companies out there
[1:00:19] are failing at this
[1:00:26] there are any other questions
[1:00:27] for Mr Montgomery
[1:00:32] right. Ryan thank you very
[1:00:34] much for your time. I don't
[1:00:36] want to run you off so you're
[1:00:38] very welcome to continue to stay
[1:00:40] and participate in the study
[1:00:42] but I've asked the honorable
[1:00:46] Mr Gum to to present a
[1:00:48] more local take on things so I'm
[1:00:50] gonna turn the floor over to him
[1:00:54] right now so thank you very much
[1:00:55] keep fighting the good fight I
[1:00:56] really appreciate you being here
[1:00:59] thank you I appreciate you guys
[1:01:01] having me and and I apologize
[1:01:02] for you know not being the
[1:01:04] greatest public speaker or I
[1:01:06] just you know like I said'm a
[1:01:08] computer nerd a computer hacker
[1:01:09] whatever you want to call me I
[1:01:10] care a ton about child safety
[1:01:12] and you I hope that I was
[1:01:14] able to explain it in a way that
[1:01:16] made sense and and understand
[1:01:17] that everyone here understands
[1:01:20] the importance of of how Ai
[1:01:22] needs to be regulated this in
[1:01:24] some way whether that be on the
[1:01:26] company level or or a legal
[1:01:26] level because kids are
[1:01:29] being hurt and and being taken
[1:01:31] advantage of and exploited and
[1:01:34] and we we as you know as
[1:01:35] people that care need to do
[1:01:36] something about it and that's a
[1:01:38] that's all I appreciate you guys
[1:01:39] hearing me out. I hope you have
[1:01:39] an awesome week.
[1:01:41] Thank you
[1:01:48] so the next presenter that we
[1:01:50] have is the honorable J Paul
[1:01:52] Gum. He is the communications
[1:01:54] director for the Oklahoma
[1:01:57] institute of Child advocacy and
[1:01:59] I really appreciate you being
[1:02:00] here today and appreciate the
[1:02:02] work that you do protecting
[1:02:04] Oklahoma's children and with
[1:02:06] that the floor is yours sir
[1:02:08] thank you senator madam chairman
[1:02:09] members thank you so much for
[1:02:10] giving OICA the opportunity to
[1:02:12] weigh in on this let me begin by
[1:02:14] giving you a bit of a history
[1:02:15] lesson about OICA
[1:02:18] we were created in 1983 in the
[1:02:19] wake of probably one of the
[1:02:21] worst child abuse scandals in
[1:02:22] state's history and that was
[1:02:26] with DHS and the problems
[1:02:27] that with children in and
[1:02:30] juvenile care the decision
[1:02:31] was made at the time to create
[1:02:33] two different entities one a
[1:02:36] state entity to operate to
[1:02:38] prevent it from inside that kind
[1:02:40] of abuse to ever occur again.
[1:02:42] OICA was created to be the
[1:02:44] outside watchdog to speak to
[1:02:45] issues
[1:02:48] of legislation of regulation and
[1:02:50] that sort of thing and we are
[1:02:51] independent we get no state
[1:02:51] money we are
[1:02:54] operated solely on donor dollars
[1:02:56] and what we can raise through
[1:02:57] our events like the heroes ball
[1:02:59] and our fall forum and the chili
[1:03:00] cook off and all the things you
[1:03:02] see us do that gives us an
[1:03:04] independence to be able to speak
[1:03:07] on issues to you and to the
[1:03:10] public on what we think is
[1:03:11] necessary to improve the quality
[1:03:12] of life for Oklahoma's children
[1:03:14] so that's where we come from and
[1:03:15] how we hope to approach
[1:03:16] things
[1:03:18] let's see here
[1:03:26] all righty Ai is a tool it's
[1:03:28] like any other tool that's out
[1:03:29] there it can be used for good or
[1:03:32] it can be used for evil it is
[1:03:36] moving at warp speed and things
[1:03:37] happen very quickly as a matter
[1:03:39] of fact the landscape changed a
[1:03:40] little bit from the time we
[1:03:43] locked thisowPoint on Thursday
[1:03:46] to today but the concerns
[1:03:48] that we have are those that
[1:03:49] we've heard shared already here
[1:03:51] today the generative image tools
[1:03:52] can take real photos and turn
[1:03:53] them into something
[1:03:54] that is just flat wrong
[1:03:57] and then social media
[1:04:00] integration is something else
[1:04:02] kidsd are on social media that
[1:04:03] the toothpaste is out of the
[1:04:04] tube they're not the kids aren't
[1:04:06] coming off social media but that
[1:04:09] is giving access to children to
[1:04:10] those who might not have the
[1:04:11] best intentions of them
[1:04:14] and existing legal frameworks
[1:04:16] designed for traditional imagery
[1:04:17] failed to address the automated
[1:04:17] generation
[1:04:20] and facilitator negligence or
[1:04:21] mal intent
[1:04:26] now our friend Mr Musk has gone
[1:04:28] front in the last 96 hours from
[1:04:30] saying it needs to continue Ai
[1:04:32] growing and expanding just
[1:04:34] exactly the way it always has to
[1:04:36] now he's beginning to suggest
[1:04:37] that maybe we need to slow
[1:04:37] things down a bit
[1:04:42] but his grok and his exAi
[1:04:44] platform are directly integrated
[1:04:46] into one of the social media
[1:04:47] platforms that's most popular
[1:04:49] and that's and that allows users
[1:04:51] to take ordinary public photos
[1:04:54] and do bad things with them
[1:04:58] so we need to make certain that
[1:05:00] criminal imagery is not being
[1:05:02] created of our children through
[1:05:04] these different platforms the
[1:05:06] Ai tool was natively embedded in
[1:05:07] the social networks so it is as
[1:05:08] much a part of it
[1:05:12] as the 140 characters you have
[1:05:14] so it's an area of particular
[1:05:15] concern
[1:05:20] Grok did acknowledge there were
[1:05:21] some problem lapses and
[1:05:22] safeguards was the euphemistic
[1:05:24] term they used so they threw in
[1:05:26] some patches to try to create a
[1:05:28] zero tolerance policy the
[1:05:30] problem is they're basic kid
[1:05:32] mode can be easily bypassed by
[1:05:36] even young people so again what
[1:05:37] we're seeing in the industry is
[1:05:38] we're not seeing enough of the
[1:05:40] safeguards being built in and so
[1:05:44] there needs to be some kind of
[1:05:45] incentive or to
[1:05:48] put the safeguards in or
[1:05:50] disincentive to not have those
[1:05:51] safeguards
[1:05:56] AP has looked into this and it
[1:05:58] sees that rok images depicting
[1:06:00] children can easily be turned
[1:06:04] into CA andXAi introduced
[1:06:05] restrictions in a Dutch court
[1:06:08] orderedAi to immediately cease
[1:06:09] generating nonconsensual
[1:06:09] undressing images
[1:06:13] jury still out on whether that
[1:06:14] was effective or not and that's
[1:06:16] just one country in an entire
[1:06:18] world and here and here's the
[1:06:19] thing about
[1:06:19] Ai
[1:06:21] and and we've heard the term
[1:06:22] the wild west used more than
[1:06:24] once this is far worse in the
[1:06:26] wild west and the wild west if
[1:06:28] you wanted to communicate with
[1:06:29] someone you either had to use a
[1:06:30] telegraph or the pony express
[1:06:34] that took time Ai is universal
[1:06:37] and it's instantaneous as soon
[1:06:40] as a tool is placed on the web
[1:06:41] it is available everywhere
[1:06:46] all the time immediately and
[1:06:47] therein creates a real problem
[1:06:49] this is far worse than what the
[1:06:50] wild west ever would be
[1:06:55] so let's look at the generated
[1:06:56] images and how prevalent it
[1:07:00] is2024 which in Ai at this point
[1:07:01] is probably ancient history
[1:07:06] a report established that 1 in10
[1:07:09] 10% of children today is aware
[1:07:11] of deep fakes and seeSA being
[1:07:11] created
[1:07:13] of their peers
[1:07:17] 1 in10 I promise you that number
[1:07:17] is higher today
[1:07:23] and then go into the chatbot
[1:07:24] therapy growing number of
[1:07:26] children and teenagers
[1:07:26] especially if they feel
[1:07:28] disconnected from the rest of
[1:07:29] the world
[1:07:30] they turned to the chatbots
[1:07:34] and they looked for advice
[1:07:36] companionship and even mental
[1:07:37] health support and again as
[1:07:37] we've done
[1:07:40] first thing you do when you try
[1:07:41] to do something with with Ai
[1:07:44] is it'll tell you Ai can make
[1:07:45] mistakes
[1:07:47] this is not an area where you
[1:07:49] want mistakes made when children
[1:07:50] are being given mental health
[1:07:51] advice
[1:07:54] and of course we've heard
[1:07:56] today the tragic results that
[1:07:57] have come from some children
[1:07:57] with this
[1:08:02] and then the severe harms of
[1:08:04] synthetic nonconsensual intimate
[1:08:06] imagery they're psychological
[1:08:09] harms it will cause
[1:08:11] psychological damage that can
[1:08:12] follow up child for the rest of
[1:08:14] their life you see extortion and
[1:08:17] coercion which we've seen that
[1:08:20] before we ever had Ai if someone
[1:08:22] could could convince a child to
[1:08:24] send an image that could be
[1:08:26] used later on to extort them or
[1:08:28] get them to get money or cause
[1:08:29] them
[1:08:30] just intentional damage
[1:08:34] and today again with the
[1:08:36] internet and with these images
[1:08:38] being instantaneous and
[1:08:40] universal the viral propagation
[1:08:43] it can go from a small town in
[1:08:44] Oklahoma to anywhere in the
[1:08:45] world
[1:08:45] like that
[1:08:46] so
[1:08:48] there are numerous
[1:08:52] cases where harm can happen that
[1:08:54] is beyond anything that it could
[1:08:55] have any time in the past
[1:09:00] so we really need to attack the
[1:09:02] criminal bad actors those are
[1:09:02] the bad guys who are really
[1:09:03] doing the the the bad things
[1:09:04] that's
[1:09:07] the statute's primarily focused
[1:09:09] on distribution of nonconsensual
[1:09:10] photos by individuals
[1:09:13] pursuing individual juvenile
[1:09:15] offenders fails to stem the
[1:09:16] automated assembly line of
[1:09:18] generating thousands of harmful
[1:09:19] images daily
[1:09:20] so
[1:09:23] some of these are being created
[1:09:25] not by people as senator
[1:09:25] Hamilton said that
[1:09:28] there's this machine in there
[1:09:31] that's making these things so I
[1:09:32] think policy needs to reach the
[1:09:34] root of the crisis and that's
[1:09:36] the commercial developers and
[1:09:38] hosting platforms that knowingly
[1:09:39] permit market or fail to
[1:09:41] implement guardrails against
[1:09:43] illicit generative sexualization
[1:09:43] of children
[1:09:47] if they can create a system
[1:09:49] whereby these images can be
[1:09:52] created and propagated they can
[1:09:53] put up the guardrails
[1:09:57] again this is one of those
[1:09:59] issues that is not partisan at
[1:10:01] all red states blue states and
[1:10:03] everything in between are trying
[1:10:05] to do things California they
[1:10:06] enacted a pioneering law
[1:10:08] requiring Ai chatbots to state
[1:10:10] that they are not human and
[1:10:11] filter explicit content for
[1:10:12] minors and spot signs of self
[1:10:16] harm and suicidal thoughts Idaho
[1:10:18] oregon Washington red blue
[1:10:20] purple passed laws stopping
[1:10:22] chatbots from acting like they
[1:10:23] have feelings are showing
[1:10:24] sentience or having sexual talk
[1:10:28] s with minors Tennessee red as
[1:10:30] Oklahoma made nonconsensual Ai
[1:10:32] generated nude images and fake
[1:10:34] explicit media illegal adding
[1:10:35] strict criminal and civil
[1:10:38] penalties and schools must also
[1:10:39] review clear rules regarding Ai
[1:10:41] use in the classroom. New York
[1:10:42] City about as blue as you can
[1:10:44] get they announced a one year
[1:10:46] moratorium on student facing
[1:10:49] generative Ai that's even using
[1:10:51] it in grades two2K through
[1:10:52] eighth grade for the upcoming
[1:10:53] school year
[1:10:58] last year Oklahoma took a shot
[1:10:59] didn't quite get there but
[1:11:02] several things were done and and
[1:11:04] I want you know give kudos to
[1:11:06] Representative conney Maynard
[1:11:08] from my hometown of Durant
[1:11:09] uhpresentative Ford Senator
[1:11:10] Hamilton and you madam chairman
[1:11:12] for the work you did on what was
[1:11:13] able to get done in last year's
[1:11:14] session
[1:11:17] butpresentative Ford had a
[1:11:18] committee substitute for Senate
[1:11:20] Bill445 last session that
[1:11:22] targeted proposals to clarify
[1:11:24] that criminal liability covers
[1:11:26] both knowingly creating and
[1:11:28] altering nonconsensual intimate
[1:11:29] images through Ai and
[1:11:29] distributing them
[1:11:32] the proposal will empower the
[1:11:33] Ag's office with direct
[1:11:34] enforcement authority while
[1:11:35] creating a civil path for
[1:11:37] victims to seek damages and
[1:11:40] mandatory takedowns of offending
[1:11:42] material those are good starting
[1:11:43] points now
[1:11:46] where we run into the problem
[1:11:48] with Ai and technology is
[1:11:48] technology doesn't have
[1:11:52] to adjourn at5 p.m. on the last
[1:11:54] Friday in May it keeps going all
[1:11:56] the time it keeps expanding all
[1:11:57] the time and that's why
[1:11:58] Senatorhamilton and this
[1:11:59] committee you're doing a good
[1:12:00] thing by continuing the
[1:12:02] discussion even during the
[1:12:04] interim but technology is
[1:12:05] advancing while we're sitting
[1:12:06] here today talking about
[1:12:07] technology advancing
[1:12:12] the key components did things in
[1:12:14] a proper way we think at OICA
[1:12:16] that protects neutral AI
[1:12:18] developers those who aren't
[1:12:19] trying to be the bad guys but
[1:12:20] again
[1:12:23] even a rock the first tool can
[1:12:26] be used to cause damage what we
[1:12:27] need to do is those who are
[1:12:28] legitimate general purpose
[1:12:30] algorithms or research models
[1:12:32] operating within safety
[1:12:32] protocols they need to be
[1:12:33] protected
[1:12:37] intentional and reckless misuse
[1:12:38] and there's where we get into
[1:12:40] the problem liability should
[1:12:42] apply specifically to those
[1:12:43] entities and actors that
[1:12:44] knowingly and recklessly design
[1:12:46] market facilitate tools for
[1:12:48] sexually explicit depiction of
[1:12:50] minors and we know that those
[1:12:52] people are out there we know
[1:12:53] that they are moving the ball
[1:12:55] down the down the field in that
[1:12:56] realm they're the targets
[1:12:58] they're the ones we should go
[1:13:00] after and then I believe civil
[1:13:02] redress and OICA does well
[1:13:04] should be part of the process
[1:13:06] you have victims and families
[1:13:06] immediate legal standing to seek
[1:13:09] compensatory damage and court
[1:13:10] ordered removal of harmful
[1:13:11] content
[1:13:14] I think that's at the foundation
[1:13:15] of everything you go to
[1:13:15] if
[1:13:19] criminal penalties aren't enough
[1:13:22] hit these people where it hurts
[1:13:23] we've heard they're making
[1:13:25] billions of dollars they are hit
[1:13:28] them in the pocketbook make it
[1:13:29] financially not
[1:13:32] worth their while to go down
[1:13:33] these dangerous roads for kids
[1:13:34] now
[1:13:38] let's take a look at what the
[1:13:39] public thinks
[1:13:42] Americans they are worried about
[1:13:43] Ai they want stronger regulation
[1:13:46] more people are worried than are
[1:13:48] excited about Ai more people
[1:13:50] believe that it is a high or
[1:13:51] very high risk than believe it's
[1:13:51] a low risk
[1:13:54] there are two areas of optimism
[1:13:56] both in scientific research and
[1:13:57] in health care and we're seeing
[1:14:00] Ai make some good strides there
[1:14:01] and that's reflected in the way
[1:14:02] people in the United States CAi
[1:14:08] but4 in5 Americans want to see
[1:14:10] more action when it comes to
[1:14:12] regulating Ai and 81% say
[1:14:14] Washington is not doing enough
[1:14:15] to regulate AI.
[1:14:19] now let's come here home to
[1:14:19] Oklahoma
[1:14:25] a 94% believe that leaders must
[1:14:25] take action
[1:14:31] 93cent are concerned about deep
[1:14:31] fakes of minor children
[1:14:36] and 91% support corporate
[1:14:36] accountability legislation
[1:14:40] this is one of those rare areas
[1:14:41] that you get as elected
[1:14:42] officials where public
[1:14:46] perception and public belief and
[1:14:48] public wants line up with what
[1:14:50] will be good policy for the
[1:14:51] state of Oklahoma
[1:14:53] it's nonpartisan
[1:14:56] people believe strongly in this
[1:14:59] you don't see 94% of the people
[1:15:00] agreeing on what kind of pizza
[1:15:02] to have for lunch but they agree
[1:15:03] that something needs to be done
[1:15:03] to protect children
[1:15:06] and there is a clear mandate
[1:15:09] that you have from the voters
[1:15:10] from the people of the state to
[1:15:11] do something
[1:15:15] so you have a rare opportunity
[1:15:16] to establish a national
[1:15:18] benchmark by closing legal
[1:15:20] loopholes around AI facilitated
[1:15:22] nonconsensual intimate imagery
[1:15:24] empowering the attorney
[1:15:24] general's office in law
[1:15:25] enforcement and protecting
[1:15:25] children
[1:15:28] basically
[1:15:29] you got the folks behind you
[1:15:33] and the time is now to take big
[1:15:35] steps and be ready to take those
[1:15:36] steps when the legislature comes
[1:15:38] back the first Monday in
[1:15:39] February
[1:15:41] with that madam chairman members
[1:15:42] I'll be happy to try to answer
[1:15:44] questions and let me conclude by
[1:15:46] extending the the regrets of
[1:15:50] Joe Dorman our CEO for
[1:15:52] not being able to be here today
[1:15:54] he's indc doing thelord's work
[1:15:56] fighting for kids in in our
[1:15:58] nation's capital so he would
[1:15:59] have rather been here than me
[1:16:02] but I tried to do ok as far
[1:16:03] as the B team goes and they'd be
[1:16:03] happy to try to answer questions
[1:16:07] any questions for the
[1:16:07] honorablein Gumm
[1:16:14] if there's none right now then
[1:16:17] I've got one or two of the first
[1:16:17] was
[1:16:21] if you might recommend a
[1:16:22] mechanism
[1:16:26] just in a I know we're kind of
[1:16:28] taking a shot in the dark here
[1:16:30] but a mechanism by which we
[1:16:34] might keep Oklahoma's kids safer
[1:16:37] and minimize the collateral
[1:16:38] damage that I think
[1:16:42] that some of the providers of
[1:16:46] the services have mentioned
[1:16:48] that would happen were we to
[1:16:50] institute some type of age
[1:16:52] verification procedures for
[1:16:54] these things I think that's
[1:16:56] critical especially when you
[1:16:57] have the Ai
[1:17:01] platforms with the ability to do
[1:17:01] the things that they do
[1:17:04] some kind of age verification
[1:17:07] it seems reasonable and
[1:17:10] appropriate and again looking
[1:17:11] back at the polling data I think
[1:17:12] you'll see Oklahomans will be
[1:17:14] supportive of that by an
[1:17:15] overwhelming majority
[1:17:19] that has always been the best
[1:17:20] way to protect children from
[1:17:22] things that might be fine for
[1:17:24] them as an adult but not fine
[1:17:25] for them as a child
[1:17:28] alcohol laws as is a perfect
[1:17:30] example of that you have to show
[1:17:32] that you're21 years old to be
[1:17:32] ordered to buy alcoholic
[1:17:33] beverages
[1:17:34] that
[1:17:37] it almost seems too simplistic
[1:17:40] but there's got to be a way to
[1:17:41] make that work
[1:17:43] thank you
[1:17:46] the the the next
[1:17:48] question I would have would be
[1:17:52] can you think of any other
[1:17:54] areas you mentioned alcohol but
[1:17:55] ones that are you know
[1:18:00] I guess that are newer that
[1:18:01] are specifically related to
[1:18:04] child safety where a person who
[1:18:06] wanted to access a service or
[1:18:10] buy a product would have to show
[1:18:12] or upload identification well
[1:18:13] just the the
[1:18:15] what the gentleman talked about
[1:18:17] with the state of Florida
[1:18:20] putting in age verifications for
[1:18:22] for online pornography and and
[1:18:23] one of the major platforms
[1:18:26] completely stopped serving that
[1:18:27] state I mean I don't know that
[1:18:28] that would be a bad thing for
[1:18:29] Oklahoma to do
[1:18:34] and I think we really have to
[1:18:36] also throw the parent in here as
[1:18:38] well the parent has to be a part
[1:18:39] of this this process I had a
[1:18:42] he's not so young anymore he's
[1:18:44] married and joining the air
[1:18:45] forcece but I have a son who
[1:18:48] grew up playing obloxx we
[1:18:50] stayed very engaged with his
[1:18:52] his mother and I stayed very
[1:18:53] engaged with his online activity
[1:18:54] a matter of fact he's 20 years
[1:18:55] old and
[1:18:58] took the oath to be in the air
[1:18:59] force and I still get notices of
[1:19:02] when he when he buys a game
[1:19:03] he hadn't figured that one out
[1:19:06] yet so I I think that that's
[1:19:08] that's part of the thing too is
[1:19:10] we need to encourage parents and
[1:19:11] empower parents to be a part of
[1:19:12] that process as well
[1:19:14] no one is better
[1:19:17] equipped to take care of their
[1:19:18] kids and their parents as a
[1:19:20] general rule and they they are
[1:19:22] that first line of defense but
[1:19:24] what we're seeing and especially
[1:19:26] with the chatbox stuff or the
[1:19:27] kids are bypassing the parents
[1:19:28] to get some kind of
[1:19:32] insight that they're not getting
[1:19:34] anywhere else so you have to
[1:19:36] find a way to short circuit that
[1:19:39] bypass and and again either an
[1:19:40] incentive to do the right thing
[1:19:42] or a disincentive to do the bad
[1:19:43] thing is's what we need to do
[1:19:43] with these companies.
[1:19:48] very well said I appreciate
[1:19:50] what you said about the you know
[1:19:51] that's the parent's job and
[1:19:52] nobody knows a kid better than
[1:19:56] the parent the you touched
[1:19:58] on this but the prevalence of
[1:19:59] technology and I'm not sure if
[1:20:01] it's what I'm saying is a
[1:20:02] question or if it's more of a
[1:20:03] statement but the the the
[1:20:07] it becoming more and more
[1:20:10] technologically savvy. I had a
[1:20:11] friend a family friend and a
[1:20:13] couple of years ago who had a
[1:20:16] child that some of the
[1:20:20] child's middle school classmates
[1:20:22] created a fake social media
[1:20:26] account for the child and then
[1:20:28] made fake posts that were
[1:20:30] running down one of the child's
[1:20:33] classmates and that resulted in
[1:20:34] an altercation in both children
[1:20:38] getting in trouble at school and
[1:20:40] so this was this was before
[1:20:42] really before the days of Ai and
[1:20:44] and I think that and you
[1:20:46] can feel free to dovetail on
[1:20:48] this or or dog pile on and do
[1:20:49] whatever the appropriate
[1:20:52] terminology is but this was this
[1:20:54] was in the days of foot boats
[1:20:56] when this was happening that
[1:20:58] junior high school children were
[1:21:00] doing that were you do would you
[1:21:02] like to expound on that or well
[1:21:04] let me say this
[1:21:08] that is not a new kind of thing
[1:21:10] the platform is new using social
[1:21:11] media but
[1:21:14] I'm not quite so old that I
[1:21:16] don't remember being in middle
[1:21:18] school and how the rumor gets
[1:21:19] started and
[1:21:22] you say one thing about somebody
[1:21:24] here and by the time it gets to
[1:21:25] that somebody there's going to
[1:21:28] be a fight after school so the
[1:21:30] platform is new the propagation
[1:21:32] is new the ease with which these
[1:21:34] things can happen is new and
[1:21:36] and again that's probably a good
[1:21:38] reason why you want to limit the
[1:21:40] use of AI in a classroom but
[1:21:42] you know it it
[1:21:45] if you've got the situation
[1:21:46] where you give the parents or
[1:21:48] you give the victims the ability
[1:21:50] to order something to be brought
[1:21:52] down to have it brought down to
[1:21:54] have monetary damages for when
[1:21:56] somebody does something like
[1:21:58] that that seems to make sense
[1:22:00] and again
[1:22:03] none of this is
[1:22:04] new
[1:22:08] it's the platform it's how fast
[1:22:11] it can get out and how fast it
[1:22:14] can go around is new and that
[1:22:16] creates the new risks for our
[1:22:18] kids and why I think it's
[1:22:20] appropriate and why most
[1:22:22] Americans and most Oklahomans
[1:22:23] think some sort of action is
[1:22:23] appropriate
[1:22:28] thank you very much. Any
[1:22:29] additional questions
[1:22:30] senator
[1:22:32] thank you for the presentation
[1:22:36] former senator by the way I
[1:22:36] don't know if anybody else knows
[1:22:38] he was served in this body a
[1:22:40] few years ago so welcome back
[1:22:42] to the state to be changed a
[1:22:43] little bit since you were here
[1:22:43] back in the day
[1:22:46] so I know this is all about kids
[1:22:48] the study is as it should be so
[1:22:50] that's certainly important to
[1:22:52] all of us this is an issue that
[1:22:53] I mean it crosses a lot of lines
[1:22:55] I think about the older folks
[1:22:56] are hears awful stories about
[1:22:58] them losing their life savings
[1:22:59] over someone that takes
[1:22:59] advantage of them so I don't
[1:23:00] know how we
[1:23:02] don't know how we get to that
[1:23:04] part of this issue as well
[1:23:06] because it's I mean it's
[1:23:06] probably not as important but
[1:23:07] it's a very important to some of
[1:23:08] these people and that happens
[1:23:10] all the time so I don't know how
[1:23:12] we can cross over and and tackle
[1:23:13] some of these same issues at the
[1:23:13] same time but that's a
[1:23:16] that's a real concern of mine. I
[1:23:17] mean I hear about those stories
[1:23:18] all the time and it's it's
[1:23:19] really disheartening when you
[1:23:20] hear about those things but
[1:23:23] I hate it for parents these days
[1:23:25] I mean there's no book on
[1:23:26] parenting there wasn't when I
[1:23:28] was raising my kids and there
[1:23:29] wasn't for my parents but there
[1:23:30] certainly isn't today but some
[1:23:32] of these issues that they're
[1:23:34] faced with I don't know how we
[1:23:35] get word to parents I don't know
[1:23:36] whether it's through the schools
[1:23:37] I don't know if the government's
[1:23:38] always the right way to do that
[1:23:40] but just to make them aware of
[1:23:41] what their kids are going
[1:23:42] through today and and again I
[1:23:44] don't know if that's through
[1:23:46] presentation at schools or
[1:23:48] wish more parents could see what
[1:23:49] we're seeing today and heres
[1:23:50] some of these pitfalls that are
[1:23:51] out there
[1:23:53] facing their kids today probably
[1:23:54] right under the nose of some
[1:23:55] parents are gonna know about it
[1:23:56] and some aren't so I don't know
[1:23:58] how we just get the word out I
[1:23:59] mean you know we can have
[1:24:01] penalties but just just to make
[1:24:03] parents aware of the situation
[1:24:05] as it exists and if this might
[1:24:06] have been a better question for
[1:24:08] Mr Montgomery before but if he
[1:24:10] were to go back ive years ago,
[1:24:12] would he have any idea that
[1:24:14] in2026 we're dealing with some
[1:24:15] of the issues we're dealing with
[1:24:17] today and then what's that going
[1:24:19] to advance to a year from now or
[1:24:20] ive years from now how do we
[1:24:21] foresee that
[1:24:23] how do we protect our not just
[1:24:24] our kids but our population from
[1:24:26] some of the things that were
[1:24:28] being inundated with so there
[1:24:29] you go. jump in and tackle any
[1:24:30] of that that you'd like to
[1:24:33] weirdoICA are working really
[1:24:34] hard to help get the word out
[1:24:38] about both the advantages of and
[1:24:39] the dangers of Ai
[1:24:42] Mr Dorman writes a column every
[1:24:43] week that's printed in a lot of
[1:24:44] newspapers across the state and
[1:24:46] we put it up online and we we
[1:24:48] get it out of pretty well he's
[1:24:50] done two or three on this
[1:24:52] subject already and so we're
[1:24:53] going to continue to do that as
[1:24:56] well and and always we we
[1:24:58] encourage and ask members of the
[1:24:59] legislature to use us as a
[1:24:59] resource as well
[1:25:04] we we do have a very brag
[1:25:06] about our we've got a good
[1:25:07] communications operation we get
[1:25:08] the word out and we're going to
[1:25:10] continue to do that whether
[1:25:13] it be on on whatever issue that
[1:25:14] helps children that has the
[1:25:16] potential to make life better
[1:25:17] for Oklahoma's children that
[1:25:18] that's our whole purpose for
[1:25:21] being and to be honest we'd
[1:25:22] love a day we're we're not
[1:25:24] needed anymore so we're going
[1:25:26] to continue to do that giving
[1:25:29] parents resources as to what is
[1:25:29] available out there but
[1:25:32] let me give you a little hope
[1:25:33] senator
[1:25:36] when you look at those numbers
[1:25:38] of the polling and and where the
[1:25:39] people are on these issues
[1:25:42] they're aware they know that
[1:25:42] there are problems and
[1:25:46] this gives you and your
[1:25:48] colleagues and extraordinary
[1:25:50] opportunity to to get out over
[1:25:52] the tip of the skis as they like
[1:25:54] to say right now I think Ai
[1:25:56] is out over the tip of the skis
[1:25:58] they're they're beyond the wild
[1:26:00] west there it's advancing so
[1:26:02] rapidly I don't think that that
[1:26:04] ive years ago anybody would have
[1:26:05] imagined this who knows what
[1:26:07] it's gonna look like in a year I
[1:26:08] you know I'm at the age now
[1:26:09] where I'm thinking more about
[1:26:10] grandkids than I am about my
[1:26:12] kids and and of course you
[1:26:13] know
[1:26:16] I'll just tell you this my son
[1:26:17] grew up online
[1:26:20] we kept very close eye on him
[1:26:21] and what he wants to do in the
[1:26:23] air force is cybersecurity that
[1:26:23] that kind of
[1:26:27] set in his head what where he
[1:26:29] wanted to take his career and so
[1:26:30] that's what he's gonna most
[1:26:32] likely be doing in the air force
[1:26:34] is working cybersecurity and so
[1:26:35] there are people
[1:26:38] who understand the challenges
[1:26:39] and what we need to do is
[1:26:42] continue to empower them and
[1:26:44] one way to do that is continue
[1:26:46] work by this legislature
[1:26:48] continued work by organizations
[1:26:50] like us to let people know what
[1:26:52] is going on and that there is
[1:26:53] hope
[1:26:56] it's not going to be the easiest
[1:26:57] thing in the world by any
[1:26:58] stretch of the the imagination
[1:27:00] but I do believe we have that
[1:27:01] opportunity
[1:27:01] before us
[1:27:08] Senator Hicks thank you well
[1:27:09] before I turn it back over to
[1:27:11] you madam chairir I got just a
[1:27:13] couple of things one of the
[1:27:16] sadly one of the things that we
[1:27:17] often ask in Oklahoma
[1:27:20] before we run any piece of
[1:27:22] legislation as what other states
[1:27:23] have done this before
[1:27:24] and
[1:27:31] so states that have passed Ai
[1:27:32] safety legislation and that term
[1:27:36] is just as broad as it can be
[1:27:37] or colo coloado,
[1:27:40] New York, Texas, Connecticut,
[1:27:41] Washington state
[1:27:43] Georgiailllinois Missouri and
[1:27:45] Michigan's got some pending
[1:27:46] Ryan Montgomery talked about
[1:27:48] Tennessee and some others that
[1:27:50] are up there so it's not an
[1:27:52] all encompassing list but there
[1:27:54] are states that are taking
[1:27:54] actions and
[1:27:58] are those actions big enough
[1:28:00] are they appropriate? Well, God
[1:28:02] will and it will be
[1:28:04] the honorable Mr Gum
[1:28:06] mentioned something about
[1:28:08] parents need to be parents and
[1:28:09] that's absolutely right
[1:28:10] I don't know what came first
[1:28:12] week in chicken or egg this
[1:28:14] thing all day long but the
[1:28:15] reason that we have kids that
[1:28:17] are doing this stuff is because
[1:28:20] we have parents that for for for
[1:28:21] for ill or good
[1:28:22] have
[1:28:26] and I don't mean to say anything
[1:28:28] bad about body because if it is
[1:28:30] but for the grace of God goi but
[1:28:32] we've got to take more of an
[1:28:34] interest in our children
[1:28:35] we've got to take more of an
[1:28:38] interest in each in each other
[1:28:40] because the sad reality is is
[1:28:42] that we we want to blame things
[1:28:44] on Ai and social media and those
[1:28:45] are those are the flavor of the
[1:28:46] month but before that it was
[1:28:48] television before that it was
[1:28:50] radio. those things are made to
[1:28:52] be more interesting than real
[1:28:53] human beings are
[1:28:57] and the it's a it's a trick they
[1:28:58] seem that way but they're really
[1:29:00] not more interesting than we are
[1:29:02] to each other they're not more
[1:29:04] interesting they're not more
[1:29:06] engaging than our children are
[1:29:08] and so if we'll do what we're
[1:29:09] supposed to do then we'll find
[1:29:11] out that our children really are
[1:29:13] just as wealth of entertainment
[1:29:13] of knowledge of
[1:29:18] ability to stay engaged and
[1:29:20] that's that's one of our most
[1:29:22] important jobs that we that that
[1:29:24] we have is to raise them up and
[1:29:25] to bring them to the age of
[1:29:28] majority physically and mentally
[1:29:30] and spiritually and emotionally
[1:29:32] intact and I saw a quote
[1:29:35] recently that I think is
[1:29:36] absolutely apropos for what
[1:29:38] we're doing here or what we're
[1:29:40] trying to to do and senator Ted
[1:29:42] Lie or I'm sorrypresentative Cha
[1:29:45] Lou the the California Democrat
[1:29:47] what he said is this we are
[1:29:49] moving from Ai that answers
[1:29:52] questions to Ai that takes
[1:29:55] actions we humans are the ones
[1:29:55] that take actions the the
[1:29:58] bus stops with us to quote Harry
[1:30:00] Truman we are the ones that are
[1:30:02] supposed to take the actions we
[1:30:03] are the ones that bear the
[1:30:04] responsibility for this we don't
[1:30:06] get the the ability we don't get
[1:30:08] the the mulligan to be able to
[1:30:10] point our finger at the computer
[1:30:11] and say it was his fault or her
[1:30:12] fault
[1:30:13] and
[1:30:16] regarding us being able to do
[1:30:19] things and take actions on
[1:30:23] various arenas that that
[1:30:23] have to do with child safety
[1:30:27] you know we can see this in
[1:30:29] that that things may have
[1:30:31] gone too far in some areas of
[1:30:33] for instance we we talked
[1:30:35] about the wild west you
[1:30:36] know in the wild west you could
[1:30:38] go down to your corner drugstore
[1:30:40] and buy a bottle of opium called
[1:30:42] laudanum. OK anybody could buy
[1:30:43] it and you can send your six
[1:30:44] year old down there with a
[1:30:46] nickel and get it or whatever
[1:30:48] and now we can't do that you
[1:30:49] can't get those drugs without a
[1:30:50] prescription, OK and that's a
[1:30:53] good thing as the FDA always
[1:30:53] done everything correct? no
[1:30:58] but have they done things that
[1:30:59] they need to do? yes they have
[1:31:01] done that we have laws that say
[1:31:02] you can't buy raw milk laws that
[1:31:04] say you can't buy meat that
[1:31:06] doesn't that hasn't been
[1:31:08] inspected by somebody and so the
[1:31:10] idea that we would not be able
[1:31:13] to pass laws that would protect
[1:31:15] mainly children but all of our
[1:31:16] citizens from some of the
[1:31:18] harmful things that come to us
[1:31:20] through the wonderful tool of
[1:31:21] the internet
[1:31:23] as I feel is errone
[1:31:26] ne ous we absolutely not only
[1:31:28] have the authority to do so but
[1:31:31] we have the duty to do so we've
[1:31:33] passed bill passed legislation
[1:31:35] regarding automobile safety
[1:31:37] firearm safety medicine safety
[1:31:38] that's the whole reason we have
[1:31:41] child safety caps on on Motrin
[1:31:42] and even on things like
[1:31:44] flintstones chewables you can't
[1:31:46] buy spray paint in Oklahoma
[1:31:48] until you turn 18. so we've got
[1:31:50] a lot of places we've got a lot
[1:31:52] of precedent where we have and
[1:31:53] we should take action on those
[1:31:54] things
[1:31:55] and I'll close with this
[1:31:56] in government
[1:31:58] we get more
[1:32:04] what we incentivize and less of
[1:32:06] what we punish and so we need to
[1:32:08] figure out what those things are
[1:32:09] that we want
[1:32:13] and incentivize those things and
[1:32:14] find out what those things are
[1:32:15] that we don't want
[1:32:18] punish those things so that we
[1:32:19] don't have any more children
[1:32:22] that have made the unfortunate
[1:32:24] decision to take their own lives
[1:32:25] we don't have people that are
[1:32:26] being damaged because they're
[1:32:28] engaging with a machine instead
[1:32:29] of with each other
[1:32:32] and I thank the staff for all
[1:32:33] the hard work that they've done
[1:32:34] I thank you madam chair for
[1:32:36] entertaining this study and I
[1:32:37] thank God that we live in a
[1:32:38] country like America where we're
[1:32:39] free to speak our minds thank
[1:32:40] you
[1:32:45] thank you senatorhamilton I
[1:32:47] probably worked on fifenate bill
[1:32:49] 1521 which is your bill
[1:32:53] more times than I worked on
[1:32:54] any of my bills list last
[1:32:55] session
[1:32:57] but I appreciate your passion
[1:32:58] and I know we will continue to
[1:33:02] work on this issue my the
[1:33:04] challenge that we continue to
[1:33:05] face is and it's in the first
[1:33:07] line of OICA's presentation is
[1:33:08] artificial intelligence offers
[1:33:11] substantial economic benefits
[1:33:12] and so we're not trying to
[1:33:14] regulate Ai comprehensively
[1:33:17] we're trying to protect our our
[1:33:18] students and our minors and our
[1:33:19] our most vulnerable Oklahomans
[1:33:20] and
[1:33:24] so my charge to industry and
[1:33:26] to anyone who feels like any
[1:33:30] of these proposed like pieces
[1:33:32] of legislation affects could
[1:33:34] affect your business and I want
[1:33:36] my Spotify to keep sending me
[1:33:37] songs that I like. I want
[1:33:38] DoorDash to keep saying hey do
[1:33:40] you also want me to deliver this
[1:33:43] food to you as well so those are
[1:33:45] the things that I want to
[1:33:47] continue and we obviously
[1:33:49] wanna protect our kids so my
[1:33:49] challenge to you is if
[1:33:53] don't come to us in February,
[1:33:54] March, April, May complaining
[1:33:56] about legislation or complaining
[1:33:58] about words your time to work
[1:33:59] with us is now to work with
[1:34:00] Senatorhamilton to work with
[1:34:02] myself. let's get this right. I
[1:34:04] think we all want to do it but
[1:34:06] this language obviously is very
[1:34:08] technical which is why it took
[1:34:10] us so much time in the spring
[1:34:12] and it got bogged down and
[1:34:14] that won't be happening next
[1:34:16] year so thank you
[1:34:16] Senatorhamilton for being a
[1:34:18] bulldog on this. Anybody else
[1:34:19] have anything else with that
[1:34:21] we're adjourned