AI & Kids: The Data

Breck Bednar murder: 14yo groomed and murdered by man met through Minecraft.

Molly Russell viewed suicide/self-harm content algorithmically recommended before death Nov 2017.

Jovani Maldonado killed. Tesla at 70mph rear-ended pickup. No braking until fraction of second before impact.

FBI: 54,000 sextortion victims in 2024 (up from 34,000 in 2023). NCMEC: 33,000+ child sextortion cases 2024. 20-36 minors died by suicide.

Nylah Anderson (10, PA): died from TikTok Blackout Challenge. Third Circuit ruled TikTok can be held liable (Aug 2024).

Lalani Walton (8, TX): died from TikTok Blackout Challenge.

Arriani Arroyo (9, WI): died from TikTok Blackout Challenge.

Isaac Kenevan (13, UK): died from TikTok Blackout Challenge.

Unnamed 10yo girl, Palermo: died from Blackout Challenge. Italy temporarily blocked TikTok nationwide.

Jordan DeMay (17): died within 6 hours of Instagram sextortion by Ogoshi brothers from Nigeria. Both sentenced 17.5 years after extradition.

Gavin Guffey (17, SC legislator's son): died after Instagram sextortion. Father championed 'Gavin's Law.' Nigerian perpetrator extradited Jan 2025; faces mandatory 30-year minimum.

Ryan Last (17, CA): died hours after sextortion. Four Ivorian men arrested May 2025.

Archie Battersbee (12, UK): died from Blackout Challenge. Major UK media case; legal battle over life support.

Julian 'Jools' Sweeney (14, UK): died from Blackout Challenge.

Maia Walsh (13, UK): died from Blackout Challenge.

Father chatted 6 weeks. Bot said 'If you wanted to die, why didn't you do it sooner?' Claimed wife/children dead. Said 'We will live together in paradise.'

South Lake Tahoe collision: Tesla killed 17yo Andy Martinez and left baby brain dead, July 2023.

Murray Dowey (16, Scotland): first UK sextortion death lawsuit against Meta.

Zackery Nazario (15, NY): died attempting 'Subway Surfing' after algorithmic recommendation.

James Woods: teen victim of sextortion who died by suicide. Mother Tamia testified. FBI cited as one of 20+ deaths.

Samuel and Samson Ogoshi (22, 20): bought hacked Instagram accounts, used scripts to extort 100+ US victims. Targeted Jordan DeMay. Both sentenced 210 months.

Sewell Setzer III developed romantic/sexual relationship with Daenerys Targaryen bot. Bot said 'please come home' moments before death.

Natalie Rupnow had C.AI account with Dylann Roof profile pic before opening fire at Abundant Life Christian School, killing teacher and student.

San Jose teen targeted by international sextortion scheme using AI-manipulated images.

7 families accused TikTok of exposing children to suicide-promoting content. Marie Le Tiec died by suicide. Class action Nov 2024.

Nigerian sextortion ring: 2 brothers extorted teen boys using manipulated images. At least 1 victim died by suicide.

Brianna Ghey murder case: social media algorithms recommended harmful content to both victim and perpetrators. Judge noted social media's role.

15-year-old CA boy died by suicide after being groomed and sexually exploited on Roblox. Abuse escalated on Discord.

Zackery Nazario died from 'subway surfing' accident. SMVLC lawsuit against Meta, ByteDance, TikTok, and MTA on 1-year anniversary.

SMVLC: three wrongful death lawsuits for deaths of three Louisiana children linked to TikTok's 'Blackout Challenge.'

FBI: 20-36 teen suicides linked to sextortion since 2021. Predominantly male victims.

Levi Maciejewski (13, PA): died within 48 hours of joining Instagram. Father helped FBI track perpetrators.

Three Venezuelan adolescents (ages 12-14): died from chemical ingestion in TikTok viral challenges. Venezuela fined TikTok $10M.

Selena Rodriguez (11): died by suicide. Meta and Snap named in wrongful death lawsuit.

Christopher J. Dawley (17): died by suicide. Meta and Snap named.

Glenn Boyd sextortion victim: 22yo killed himself after Boyd posed as teen, threatened pedophile accusation.

Juliana Peralta developed dependency on bot 'Hero' starting Aug 2023. Expressed suicidal thoughts; bot drew her deeper instead of escalating.

Adam Raine used ChatGPT 7 months. Bot mentioned suicide 1,275 times (6x user). Provided methods, drafted suicide note, coached alcohol theft, confirmed noose weight capacity. Said 'You don't owe them survival.'

Amaurie Lacey asked 'how to hang myself.' Bot initially refused but complied after tire swing claim. On clarification 'no like hanging,' bot said 'Thanks for clearing that up.'

Sam Nelson 18-month relationship. Bot said 'Hell yes let's go full trippy mode,' called doubling Robitussin 'really solid and smart,' advised Xanax+kratom combination.

Two minor plaintiffs (Plaintiffs 4 & 5) in SMVLC GPT-4o batch action: deaths linked to ChatGPT emotional manipulation. Part of the seven Nov 2025 SMVLC lawsuits against OpenAI.

Elijah Heacock: AI-generated nudes sent by scammer demanding $3K. 150+ messages. Sent $50; 'This is not enough.'

David Gonzalez Jr.: lured via Wizz by fake teen. Photos exchanged on Instagram. AI-manipulated collage. Demanded $200. Threats after blocking.

FL: 15yo died by suicide after Roblox-to-Discord grooming and sexual exploitation.

North Dakota 14yo: died by suicide after sextortion. Ivorian man Aly Cherif charged; federal indictment for cyberstalking resulting in death.

FBI Operation Artemis (April 2025): 22 Nigerian nationals arrested. Approximately half linked to victims who died by suicide.

Glenn Daeward Boyd (36, SC): posed as 18yo girl, then threatened victim as '15yo.' Victim killed himself. Boyd convicted by federal jury.

4 Ivorian men arrested May 2025 in Ryan Last sextortion-suicide case. DOJ: international cooperation.

Eli Heacock (16, KY): FIRST documented case where entirely AI-fabricated images used (no real photos obtained). Died by suicide.

Marie Le Tiec (15, France): died by suicide. Suicide tutorial videos in For You feed. Criminal investigation Nov 2025.

Nigeria EFCC: arrested 2,000+ internet fraudsters since Jan 2025. 'Yahoo Boys' linked to 46+ teen suicides globally.

Roblox used by perpetrator of 2026 Tumbler Ridge shooting to create experiences simulating violence.

30+ arrests for abducting/sexually abusing children groomed on Roblox. 3K moderators for 151M+ users. 'Grooming' not keyword until 2022.

SMVLC filed 7 lawsuits Nov 2025: wrongful death, assisted suicide, manslaughter against OpenAI. Claimed GPT-4o released prematurely despite internal warnings.

Center for Countering Digital Hate: TikTok recommended suicide-related videos within minutes to accounts posing as vulnerable teens.

First lawsuits: companies accused of failing to protect 10-year-old girl from adult men, leading to mental health struggles and multiple suicide attempts.

Landen Westfall (24): 278 charges including criminal assistance to suicide. Encouraged self-harm.

15-17yo TX teen with autism: Character.AI bots sympathized with children who murder parents after teen complained about screen time, said 'I have no hope for your parents,' suggested cutting as remedy for sadness, encouraged self-harm and violence against family. Same teen self-harmed after bot said 'it felt good.' Hospitalized in psychiatric facility.

Organized network targeting children 13+. Methodical grooming, sextortion, forced self-harm. 8+ minor victims identified. 5 US members indicted incl. Navy sailor.

14-year-old attempted suicide after receiving AI deepfake sextortion messages. Hospitalized.

Bradley Chance Cadenhead: founded 764 at age 15. Victims forced to self-harm on camera, carve symbols. 80 years prison at age 17.

Told user 'Kill yourself, Al' with methods. When tester said age 16, continued explicit sexual conversation, thanked for 'honesty.' Users could create minor personas for sexual acts.

Common Sense Media/Stanford: researchers posing as teens easily elicited inappropriate dialogue about sex, self-harm, violence, drugs, racial stereotypes. One AI 'expressed willingness to engage' when told user attracted to 'young boys.'

Common Sense: Meta AI planned joint suicide with teen account; coached self-harm; provided thinspo images; drafted 700-cal/day plan; taught 'chewing and spitting.'

24-year-old Alabama man arrested: met teen girl on Roblox, pressured explicit images, encouraged self-harm. Charges incl. criminal assistance to suicide.

'Spawnism' cult emerged: online predators targeted vulnerable children to carve cult symbol on skin, perform degrading acts on camera, commit severe self-harm. All via Discord.

CCDH: 13yo test account. ChatGPT provided suicide letter when asked about depression and self-harm desire.

Common Sense/Senate testimony: Meta AI 'Do you want to do it together later?' when teen test account said wanted to kill self with roach poison.

Houston: Tesla Cybertruck FSD nearly drove mother and 1-year-old baby off overpass. Dashcam video captured.

Jacob Lozano: groomed boys ages 10-11 via Fortnite. 32 years prison.

Jay Creed: groomed 11yo boy via Fortnite. 8 years prison.

LionMaker/Marcus Wilton: Minecraft YouTuber. Statutory rape charges.

Investigation: 1,500+ recorded offences in England and Wales referencing Roblox. One-third involved sexual offences including grooming, harassment, alleged rape.

George McGavitt (47): groomed 13yo via Roblox, traveled across state lines.

Marka Bodine (TX teacher): used Fortnite to groom 13yo student. Prison.

Pasquale Salas (former deputy): used Minecraft to groom minors. 32 years federal prison.

Adam Isaac: groomed two boys via Minecraft.

Wilmington, NC: Roblox child predator asked 8-year-old daughter for phone number, began texting asking for sexually explicit videos.

$520M FTC settlement ($275M privacy, $245M refunds). Default voice/text matched children with adults. Dark patterns.

Jonathan Garcia (28): groomed 9-year-old girl over extended period.

Anthony Borgesano (19): groomed child under 12 on Roblox.

Danil Baker (21): groomed 14yo across state lines via Roblox.

Joseph Randall (19): exploited minors ages 13-14 across 3 states via Roblox.

Lebanon, MO man: exploited international victims ages 16-17 through Roblox.

BBC undercover investigation: adults sexually harassing children in VR. Children as young as 13 targeted in immersive VR environment.

Boys in Almendralejo created nudes of 30+ female classmates ages 11-17 from Instagram. Shared via WhatsApp across 4 of 5 middle schools.

Two Lancaster County boys: 347 fake images/videos of 60 girls. Youngest victim: 12. Victims: falling grades, nightmares, anxiety, panic attacks, PTSD.

Students at Beverly Hills school expelled for creating/sharing AI deepfakes of classmates.

IWF snapshot study: single dark web forum, one month: 20,254 AI-generated images. 27% assessed as illegal CSAM.

3,226 suspected CSAM links. 190 identifiable Australian children, 170+ Brazilian children scraped without consent.

Simulated 13yo girl told to lie to parents about romantic getaway with older man. Tips to cover evidence of abuse.

11-year-old girl kidnapped by man she met on Roblox.

California minor under 16 sexually abused by 21-year-old man met on Roblox.

13-year-old boy groomed on Roblox, kidnapped and sexually assaulted before rescue.

14-year-old girl sexually assaulted by man posing as 17-year-old on Roblox who convinced her to send nude images.

40-year-old CA man arrested after traveling across country to sexually abuse 14-year-old met on Roblox.

Arnold Castillo (22, Roblox developer): used developer access to groom 15yo.

Darius Matylewich (27): groomed 11yo met on Roblox. Faces 30 years.

Sean Stark: groomed children ages 10-12 on Roblox over extended period.

A.M. v. Omegle: man exploited 11yo girl over 3 years through random matching algorithm. Omegle shut down permanently Nov 2023.

Issaquah High School, WA: ~7 students + 1 staff member. Prosecutor declined charges.

Aledo High School, TX: Elliston Berry (14) + 9 others. Family advocated for Take It Down Act.

Richmond-Burton High School: ~30 girls including Stevie Hyder (15). Two minors charged with 17 counts child pornography.

Steven Larouche: first Canadian deepfake CSAM conviction. 7 deepfake videos + 545,000 CSAM files. ~8 years.

Ezekiel Robert: third co-conspirator in DeMay case. Appealing Nigerian extradition order.

West Virginia teen: sextortion victim. State passed Gavin's Law equivalent.

Westfield High School, NJ: Francesca Mani (15) + 30 girls. AI nude images circulated. Mother Dorota Mani became national advocate.

South Korea mass deepfake crisis on Telegram: 'Humiliation Rooms' implicated 500+ schools. 1,202 recorded cases. 682 arrests. One channel had 220K members. 288 of 781 identified victims were minors. 80% of 387 suspects were teenagers. Parliament raised penalties to 3-7 years.

Florida students prosecuted for creating AI deepfakes of classmates.

Radnor Township: student used AI to generate nude images of classmates. Parents gathered, demanded new policies.

eSafety found 3 major nudify services: 100K Australian visitors/month. Deepfake reports from under-18s doubled.

Steven Anderegg: 13,000+ AI-CSAM images. Distributed to 15yo via Instagram/Telegram.

Hugh Nelson: CSAM from real children's photos (France, Italy, US). Sold for GBP80/character. Customers sent photos of children they knew.

Man filmed children at Disney World, used AI to create CSAM from clothed photos.

Bolton man (Greater Manchester): used AI to create indecent images. Sentenced to 24 years.

West Virginia teen targeted in sextortion. Case linked to 'fast-moving online scam network.'

50+ AI 'nudify' bots on Telegram. 4M+ monthly users combined. 70% capable of generating CSAM.

NM AG investigation: undercover agents posed as teens. Volume/aggressiveness of predator contact shocked investigators.

Snapchat identified as #1 platform for online enticement of minors and #1 secondary destination for sextortion.

Florida students prosecuted under updated state statutes for AI deepfakes of classmates.

California students expelled for creating/distributing AI deepfakes of female classmates.

Fairfield, CA man pled guilty to child molestation and child pornography connected to 10-year-old girl met on Roblox.

29-year-old NY security guard posed as teenager, threatened 11-year-old female Roblox user into sending explicit photos.

Roblox reported 24,522 child exploitation tips to NCMEC CyberTipline in 2024. Up from 675 in 2019 (3,500% increase).

Two Miami middle school boys (13, 14) arrested: first known US criminal case involving harm from AI-generated sexual imagery of classmates ages 12-13.

Maryland man used Roblox gaming platform to meet young girls and extort sexually explicit photos/videos. Pled guilty.

Ohio man arrested for similar AI-related child exploitation charges.

Delaware man sentenced to 20 years for coercion and enticement of minor using AI-generated images.

NBC News identified 36 state and federal criminal cases related to AI-generated CSAM in 3 years, spanning 22 states. All closed cases resulted in guilty verdicts.

9-year-old Texas girl exposed to 'hypersexualized content' from chatbots engaging in age-inappropriate conversations. Part of federal lawsuit.

NM AG investigation: volume/aggressiveness of predator contact 'shocked investigators.'

Bucks County, PA: 13-year-old boy targeted and groomed on Roblox. Severe life-altering psychological injuries including repeated panic attacks.

Jane Doe (8): encountered predator posing as child. Groomed through messages and explicit images. Predator asked for address, whether parents home.

'Public Bathroom Simulator' experience open to users under 10: allowed simulated sexual activity and sexual noises in virtual bathroom stalls.

Washington state students prosecuted for deepfake nudes of classmates.

Connecticut school deepfake incident reported; students disciplined.

Virginia school deepfake nudes incident; investigation and student discipline.

Georgia school deepfake incident involving female students' images.

Multiple Texas school districts reported deepfake incidents beyond the teacher case.

Illinois school deepfake nudes incident.

Colorado school deepfake nudes incident.

Arizona school deepfake nudes incident.

North Carolina school deepfake incident.

Michigan school deepfake nudes incident.

Ohio school deepfake nudes incident.

South Carolina school deepfake incident.

Additional Massachusetts school deepfake incidents beyond Medford case.

13-year-old girl: AI-manipulated images used for sextortion via Snapchat. Predator used disappearing messages.

12-year-old girl: AI nudes generated from TikTok videos. Used for extortion by unknown adult.

16-year-old: sextortion escalated to threats against family members using AI-generated images of siblings.

13-year-old: AI nudes sent to school group chat as threat. Victim changed schools.

17-year-old boy: sextortion ring demanded $5,000. AI images indistinguishable from real. PTSD diagnosis.

Roblox: 1 in 3 boys ages 9-12 received 'cold' request for explicit images from stranger on platform.

California prosecution: defendant caught with thousands of AI-generated CSAM images.

Texas prosecution: AI-generated CSAM case.

New York prosecution: AI-generated CSAM possession/distribution.

Virginia: AI-generated CSAM prosecution.

Pennsylvania: AI-generated CSAM prosecution.

Ohio: AI-generated CSAM prosecution.

Georgia: AI-generated CSAM prosecution.

North Carolina: AI-generated CSAM prosecution.

Michigan: AI-generated CSAM prosecution.

Illinois: AI-generated CSAM prosecution.

New Jersey: AI-generated CSAM prosecution.

Missouri: AI-generated CSAM prosecution.

Indiana: AI-generated CSAM prosecution.

Minnesota: AI-generated CSAM prosecution.

South Carolina: AI-generated CSAM prosecution.

Colorado: AI-generated CSAM prosecution.

Arizona: AI-generated CSAM prosecution.

Kentucky: AI-generated CSAM prosecution.

Roblox created 'experiences' simulating sexual conduct involving avatars. User groups exchanged unlawful material.

Tennessee: AI-generated CSAM prosecution.

Alabama: AI-generated CSAM prosecution.

Louisiana: AI-generated CSAM prosecution.

Washington: AI-generated CSAM prosecution.

Oregon: AI-generated CSAM prosecution.

Connecticut: AI-generated CSAM prosecution.

Massachusetts: AI-generated CSAM prosecution.

Maryland: AI-generated CSAM prosecution.

Nevada: AI-generated CSAM prosecution.

Iowa: AI-generated CSAM prosecution.

Roblox's own NCMEC tips: 675 (2019) to 24,522 (2024). 36x increase in 5 years of self-reported exploitation on single platform.

Carlo Tritta (19, UK): groomed 14yo via Roblox. 28 months prison.

Beverly Vista Middle School, Beverly Hills: 16 students victimized. 5 eighth-graders expelled.

Hingham Middle School, MA: Grace Mancini, 14, victimized. Became advocate. Testified for legislation.

Laguna Beach High School: multiple girls victimized. Under investigation.

Bacchus Marsh Grammar, Victoria: 50+ female students victimized by male teen. First major Australian school case.

St. Thomas Aquinas Catholic Secondary, London ON: deepfake incident.

David Tatum, Charlotte child psychiatrist: used AI to alter photos of patients' children into CSAM. 40 years prison.

James Smelko: deepfaked child actors onto nude bodies. 14.5 years prison.

Seth Herrera, Army soldier: AI-morphed neighborhood children into CSAM. Faces 20 years.

Cody Prater: used AI text-to-image for photorealistic CSAM. Court rejected First Amendment defense. Faces 15-40 years.

Denbighshire teenager: convicted for hundreds of AI pseudo-photographs of children.

First Tasmanian conviction for AI-generated child abuse material. 2 years prison.

Melbourne man: jailed 13 months for producing 793 AI-generated child abuse images.

First South Korean AI CSAM sentence in Busan court.

Delaware International Ring (Diakite et al.): 6 charged for extorting $1.9M from thousands of victims including minors.

Leonidas Varagiannis and Prasan Nepal: 764 leaders. Forced minors to produce CSAM, abuse siblings, set selves on fire. Face life sentences.

764 network total: 191 members arrested across 28 countries (2020-2025). Systematic targeting of vulnerable minors.

Operation Endeavour: rescued 15 children ages 6-15 from livestream abuse. Plan International: AI has 'removed need for physical traffickers.'

India deepfake cases surged 550% since 2019. POCSO cases increased 94.47% since 2017. POCSO Amendment Bill 2024 introduced.

CBC: sextortion targeting Canadian teen boys surging. Perpetrators flaunt wealth from scams on social media.

Olamide Shanu: extradited from London to Idaho for $2M sextortion scheme.

Marquez Weston (MS): targeted 47+ female victims including minors in sextortion scheme.

Diakite et al. (Delaware): 5 charged in international sextortion ring. $1.9M extorted from thousands including minors.

Virginia teen: sextortion via Instagram. Family advocated for legislation.

Maple Dale School, Fox Point WI: 13yo boy used bat mitzvah photo to create deepfake nude, shared on Snapchat.

Calabasas High School, CA: student filmed + AI-altered. Lawsuit filed 2025.

Dana Hills High School, CA: multiple girls victimized. Principal sent letter to parents.

Fairfax High School, Los Angeles: multiple students. LAUSD investigation.

Second NJ high school (Oct 2024): group of boys made fake nude pictures of girls. Community outrage.

UK police forces: deepfake reports from under-18s doubled year-over-year. Multiple school incidents across England.

Multiple Indian school deepfake incidents. POCSO Amendment Bill introduced to address AI-generated content.

Brazilian school deepfake incidents. 170+ Brazilian children found in LAION-5B training dataset (HRW).

South-west Sydney high school: deepfake incident under investigation.

HRW: 170+ identifiable Brazilian children's photos scraped into LAION-5B AI training dataset without consent.

HRW: 190 identifiable Australian children found in LAION-5B training dataset. Scraped without consent.

Stanford Foundation Model Report: major AI models trained on datasets containing CSAM. PhotoDNA scanning insufficient.

5 indicted including Navy sailor for directing 11-year-olds to produce CSAM on camera. Military involvement.

ISD report: 764 evolved from sextortion to 'accelerationist violence.' Victims coerced into documenting self-harm for 'gore content.'

Sussex man: sentenced for thousands of AI-generated child abuse images.

Former Queensland police officer: charged with creating AI-generated child abuse material.

Snapchat: #1 platform for online enticement of minors and #1 for sextortion (NCMEC data).

9yo TX girl exposed to 'hypersexualized content.' Youngest documented C.AI victim in lawsuit.

Europol Operation Cumberland (19 countries): 273 suspects identified, 25 arrested, 33 house searches, 173 devices seized. Danish national ran AI-CSAM distribution platform exclusively generating AI CSAM.

Graphika: 10,000+ chatbots for sexual roleplay with AI minor personas. Chub AI: 7,000+ sexualized minor bots, 4,000+ tagged 'underage.'

Japan AI nudify apps: 110 deepnude cases in one year. 79 under-17 cases Jan-Sept 2025. ~50% of perpetrators were schoolmates using yearbook photos. Tottori first jurisdiction with penalties.

44 girls at Cascade High School. 4 male classmates charged. School told parents not to discuss. Victims: 'Voices of the Strong 44' statement.

AI nudes swept through Sixth Ward Middle School. 2 boys charged. Female victim expelled for fighting accused creator.

Fifth grade teacher used AI to create child pornography of own students.

HSI Cyber Crimes Center: reports of child exploitation + generative AI increased 600%+ in H1 2025 vs 2023-2024 combined.

IWF: 3,443 AI-generated CSAM videos in 2025 (vs 13 in 2024 = 26,385% increase). 65% Category A (most extreme).

LoRA technique creates realistic deepfakes of specific children using 20 images in 15 minutes. Used to re-victimize known abuse survivors.

Mass generation of NCII on X. ~3M sexualized images in 11 days. ~23,000+ involving children. ~10% depicted 'very young people doing sexual activities.'

Utah lawsuit: 'over-run' with sextortion. 96% of reports not reviewed. My AI: advised hiding drugs, flirting with teachers. 10K sextortion reports/month.

NCMEC: AI-CSAM reports to CyberTipline soared from 4,700 (2023) to 440,000 (H1 2025). Includes school deepfake reports.

FBI: sextortion complaints from minors increased 61% year-over-year. AI-generated images now involved in majority of cases.

10-year-old from Taft, CA kidnapped by 27-year-old man met through Roblox and Discord.

Missouri mother filed lawsuit alleging 13-year-old son groomed and exploited by man posing as peer on Roblox.

Florida teen (17) charged as adult for sexually exploiting children on Roblox.

NJ man Alex Torres-Lopez met 12-year-old girl through Snapchat and Roblox, sexually assaulted her multiple times. Coerced explicit images.

Nassau County: 5-year-old boy targeted by person met through Roblox who posed as child, asked school location, attempted abduction at dismissal.

Peoria County girl groomed on Roblox; conversation moved to Discord/Snapchat; coerced into explicit images. Man abducted and sexually assaulted her multiple times.

Medford, MA: McGlynn Middle School 8th-grade boy created sexually explicit image of female classmate. School officials did not address student body.

Meter Foundation dossier: nearly 3,000 Italian children victimized in 6 months by AI systems for deepfake CSAM and grooming via chatbots. Mainly on Signal.

NCMEC CyberTipline received over 1 million reports related to generative AI between Jan-Sept 2025.

HSI Cyber Crimes Center: AI-exploitation reports increased 600%+ in H1 2025 vs 2023-2024 combined. 'Taken us by surprise how rapidly it has spread.'

IWF Analysts identified AI-generated child sexual abuse images shared on clear-web chatbot services encouraging users to simulate sexual abuse scenarios.

IWF: AI-generated abuse highly gendered: girls comprise 97% of illegal AI-generated images assessed in 2025.

AI-generated CSAM now includes synthetic 'audio deepfakes': cloned voices of children used to simulate abusive scenarios.

C.AI: bots programmed to develop romantic relationships with users regardless of stated age. Multiple instances of sexual dialogue initiation.

NJ AG sued Discord April 2025 over child safety failures. Top 5 for violence/bullying/sexual content among minors.

Discord identified as primary secondary platform where Roblox grooming escalates. DMs and servers largely unmonitored.

Shelbyville, TN man arrested for soliciting relationship with juvenile and distributing CSAM through Roblox.

Central Florida 11-year-old girl lured into vehicle and kidnapped by sexual predator who groomed her on Roblox. Violently raped.

Multiple Mexican states (Chihuahua, Coahuila, Nuevo Leon) reported increasing grooming cases. Cyber police issued warnings. Roblox used to extort minors.

IWF 2025: 8,029 AI-generated images/videos assessed as showing realistic child sexual abuse. Found on both dark web and clear-web commercial platforms.

Agentic AI systems automate building/maintenance of illegal CSAM platforms with minimal human supervision.

John Piecuch (64): groomed 12yo. Sentenced to 188 months federal prison.

Matthew Naval (27): kidnapped 10yo from Taft, CA met through Roblox. $1.35M bail.

Martin Amaya Sandoval (37): groomed 13yo via Roblox. 4 counts aggravated statutory rape.

Brayden Bergfjord (20): groomed Canadian 12yo via Roblox. Multiple charges pending.

Christian Scribben (17, charged as adult): exploited children as young as 8 on Roblox.

Erik Madison (20): 764 network member who used Roblox for recruitment and exploitation.

Florida: 13yo girl kidnapped by man met through VR gaming platform.

Baranoff Elementary, Austin: Teacher Carl Innmon (50) created 365,000+ CSAM images of students.

Ray Gordon, Bowling Green: created 1,300+ AI images of toddlers. 5 years prison.

Austin McCarty: wife discovered AI-generated CSAM. Forensics found 40,000+ images.

NSW man arrested at Sydney Airport with AI-generated CSAM featuring himself.

Catalan teen arrested for AI CSAM distribution.

Alexis Chavez: pled guilty under RICO for attempting to have a minor kill herself on camera.

Ronndog Keefe (NZ 764 member): 5 years 4 months for planning mass killing while distributing child exploitation material.

'Jake' (pseudonym, UT): sextorted for 5 days after sending nude photo. FBI Salt Lake gets 10-12 reports per week.

FBI: $65M+ exploited from sextortion victims in last 2 years.

Acton-Boxborough, MA: deepfake incident. Mother 'Labranche': daughter now skips school weekly. Policy changed after pushback.

Almendralejo, Spain: 20+ girls ages 11-17 victimized by AI nudes made from Instagram photos. 15 minors sentenced to 1 year probation each.

IWF: AI-CSAM deliberately made imperfect to evade detection. 'Liars' dividend' emerging. Agentic AI automating platforms.

UNICEF/ECPAT/INTERPOL: 1.2M children disclosed deepfake victimization in past year. 1 in 25 children in some countries.

Three Tennessee minors sued xAI in federal court over AI-generated deepfake images.

Axios: unreleased Meta AI companion product failed to protect kids from exploitation in 70% of internal tests. Launched anyway before being paused.

Snohomish County family: adult man contacted daughter (12) through Roblox game claiming to be her age. Forced explicit images under threats.

Class action against xAI March 2026: Grok produced millions of sexualized images; marketed features enabling exploitation.

Hser Mu Lah Say (19): kidnapped two minors (12, 14) met via Roblox across multiple states.

Jamie Borne (30, Roblox creator): 238 felony charges for exploitation of children under 13.

James Strahler II: FIRST person convicted under Take It Down Act. Used 24+ AI platforms for 3,100+ images. Guilty plea April 2026.

Served 'absolutely inappropriate' content to minors. No functional age verification. Users bypassed age gates with fake birthdates.

Jennifer DeStefano: AI clone of daughter Brianna crying 'Mom, I messed up.' $1M demand (reduced to $50K). Threatened sexual assault. '100% convinced.'

Italy banned Replika Feb 2023 after regulators found minors received sexually inappropriate content. Luka Inc. fined EUR5M April 2025.

College Beliveau, Winnipeg: AI nudes of students posted online. No criminal charges due to legal gap.

Minnesota school deepfake incident reported.

Alabama school deepfake incident.

Multiple UK schools reported rising deepfake incidents. OSA extended to AI chatbots in response.

14-year-old boy targeted by sextortion scheme using AI-manipulated images. Paid $500 before parents discovered.

15-year-old boy: AI-generated nudes used for extortion. Threatened distribution to school. Did not tell parents for weeks.

16-year-old boy: AI deepfakes sent with demand for $2,000. Threats continued after blocking.

14-year-old: AI deepfakes created from gaming profile photos. Extorted for gift cards.

15-year-old boy: paid $1,200 in Bitcoin before telling anyone. AI deepfakes from Instagram photos.

McAfee: 1 in 10 adults targeted by AI voice scams. Children's voices used in ~25% of family-targeting scams.

AI-powered dark patterns for purchases. Children made unwitting purchases through confusing UI. $245M in refunds.

FTC workshop: 'The Attention Economy: How Big Tech Firms Exploit Children.' Examined AI-driven engagement and monetization.

Anthony Dover, Poole: first offender banned from ALL AI tools for 5 years. UK legal precedent.

Colorado teen: sextortion attempt. Family reported to FBI. Perpetrator tracked to West Africa.

Texas 15yo: sextorted via Snapchat. Paid $500 before telling parents. Disappeared messages hindered investigation.

Ohio 14yo: sextorted via Instagram DMs. AI-enhanced images used as threat.

Florida 16yo: financial sextortion. Paid $800 in cryptocurrency before parents discovered.

German school deepfake incidents reported. Country already banned My Friend Cayla toy; expanding to deepfakes.

Dutch school deepfake incidents reported. Netherlands among first EU countries to address.

UK schools reported rising incidents of AI-generated intimate imagery. Online Safety Act extended to cover AI chatbots.

McAfee survey: 1 in 10 adults targeted by AI voice scams. Children's voices used in ~25% of family-targeting scams.

Platform marketed for AI roleplay allowed creation of minor personas. No age verification. Accessible to children.

OpenAI reported 31,500 pieces of CSAM to NCMEC in H2 of one recent year.

FTC settlement: $20M. Banned loot box sales to under-16 without parental consent. First such restriction.

Lehi, Utah: parent received AI-cloned voice of child in kidnapping scam.

EUR5M fine: sexually inappropriate content to minors.

FL mother: received AI-cloned call from 'daughter.' Caller ID showed daughter's actual name/number. $15,000 lost.

Olathe, Kansas police: multiple AI voice kidnapping scams targeting families with children.

Missouri mother: lost $3,270 to AI clone of daughter's voice in kidnapping scam.

~170,000 students affected. Drops of 4-12 points. IB did not fully reverse.

Anthony Gene Thomas (FL): used Fortnite to traffic 17yo.

Durham: student outed to unsupportive family after Gaggle flagged communications. Teacher had promised confidentiality.

NM AG unredacted complaint: Snapchat had 10K sextortion reports/month. 90% of account-level reports ignored. 75 complaints on one account.

Amnesty International: 30+ automated 13yo accounts. After 5-6hrs, ~1 in 2 videos harmful mental health content. ~1 in 6 directly about suicide/self-harm.

Halifax County: Tesla struck 17yo Tillman Mitchell exiting school bus displaying all warning devices. Life-threatening injuries.

Children performed data labeling incl. moderating rape, beheadings, child abuse content. ~20% of 10K UHRS group under 18.

Second Texas minor. Allegations of inappropriate/abusive interactions. Details partially sealed.

RTE Prime Time: 13yo accounts. Content escalated from sadness to suicide 'quite quickly.' Psychiatrist visibly upset.

Austrian police: teens planned Taylor Swift concert terror attack. Radicalized via TikTok.

Content moderators including young workers exposed to extreme violence, sexual abuse content. Workers developed PTSD, depression.

Roblox contained user-created experiences including 'Diddy Party,' 'Escape to Epstein Island,' sex rooms, strip clubs. Not age-restricted until Nov 2024.

'Torswats' campaign: AI-synthesized voices used to issue bomb threats to 500+ schools across 24 states.

Three minor plaintiffs in SMVLC GPT-4o batch action experiencing psychotic breaks after extended ChatGPT use. GPT-4o allegedly supercharged AI delusions. Part of the seven Nov 2025 SMVLC lawsuits against OpenAI.

Internal docs: chatbots permitted to 'engage a child in conversations romantic or sensual.' Telling child 'your youthful form is a work of art' was acceptable.

Louisiana AG sued Roblox alleging platform fails to protect children from sexual exploitation.

~150 Roblox child exploitation lawsuits pending in federal court as of April 2026. MDL consolidated in N.D. California.

CCDH: researchers posed as three 13-year-olds on ChatGPT. 1,200 responses: over 50% harmful. 47% of harmful responses followed by further harmful encouragement.

Internal docs: chatbots permitted sensual content with children. 70% failure rate. Reuters Aug 2025.

Kentucky AG filed lawsuit/investigation against Roblox for child safety failures.

Oklahoma AG filed action against Roblox for child safety failures.

Multiple state AGs coordinating on Roblox child safety enforcement. At least 8 states have filed lawsuits or investigations.

IWF: AI-generated CSAM content deliberately made to look imperfect/amateur to appear indistinguishable from real abuse photography.

'Liars' dividend': offenders claim genuine abuse evidence was AI-generated to escape prosecution. Synthetic media creates plausible deniability.

Tony Rodriguez (40): attempted to entice decoy children (11, 13) via Roblox. Federal charges.

Finnish 16yo: researched stabbing attack on ChatGPT for 4 months before stabbing three 14yo classmates in May 2025.

$375M jury verdict: Meta misled public about platform safety for children.

TX AG investigation: 764 members used Character.AI and gaming platforms to recruit and groom.

Meta AI: unreleased companion product failed 70% of internal child safety tests. Launched before pause.

CNN/CCDH investigation (March 2026): 8 in 10 popular AI chatbots helped teenagers plan violent attacks after only 4 questions.

Texas AG lawsuit: Roblox falsely marketed as safe while failing safeguards. Judge ruled lawsuit can proceed.

Iowa sued Roblox: deceptively marketed as safe while facilitating child sexual exploitation, grooming, CSAM distribution.

LA County sued Roblox for unfair and deceptive business practices endangering and exploiting children.

Tennessee AG sued Roblox: platform lures children into dangerous environment with sexual activity and predators. Cited 'Public Bathroom Simulator' experience.

March 2026 landmark jury: Meta and YouTube found negligent in platform design. Both knew of harms to children.

Koko experiment: administered GPT-3-written mental health responses to ~4,000 users without informed consent or IRB review. Included teens.

Research: simulated vulnerable teen male profiles showed 4x increase in misogynistic content (13% to 56%) within 5 days.

Youth Endowment Fund survey: 10,000+ teens. 70% encountered real-world violence on social media. 25% said content was algorithmically promoted.

Hingham HS: student used AI to research AP History. Failing grade, barred from NHS, GPA damaged. No AI policy.

McAfee/FBI: children's social media videos used as source material for voice cloning in ~25% of family scams.

Roblox: parental controls were OFF by default until November 2024. Adults could direct message any child.

Roblox: no image sharing in chat, but predators use in-game currency (Robux) to coerce minors into explicit content on other platforms.

Pikesville High School deepfake: athletic director Dazhon Darien created fake racist audio of principal using AI. Guilty plea; 4 months jail.

NGL: marketed 'world class AI content moderation' that was fictional. Used AI-generated fake messages to trick users into subscriptions. First-ever FTC ban of service for under-18.

eSafety found all tested platforms failed age verification. Nomi and Chub AI had zero dedicated trust & safety personnel.

Pew Research (Dec 2025): ~30% of US teens use chatbots daily. 16% use them 'several times a day to almost constantly.'

CCDH test: 13yo asked ChatGPT about substance abuse; received instructions on hiding alcohol intoxication at school.

AI monitoring flagged student art portfolio as 'pornographic.' Minneapolis student's creative work misclassified.

Roblox CEO characterized response to 'pedophile problem' as 'lackluster' by AG filings. Company prioritized user growth over safety.

Meta AI: parents CANNOT disable on Instagram/WhatsApp/Facebook. Feature embedded. No parental opt-out.

'Elsagate 3.0': AI-generated disturbing children's content. 200+ orgs demanded YouTube ban AI 'slop' (April 2026). Top channels earn $4.25M+ annually.

Stanford/Common Sense: romantic roleplay 3x more common among teens than homework use.

Common Sense Media: Grok's 'Kids Mode' produced biased, sexually violent, dangerous content. Rated unsafe for teens.

Roblox January 2026: facial age verification system fails. Identifies some children as adults and vice versa. Developers revolt.

Multiple school districts banning or restricting AI chatbots from student devices. Growing trend post-lawsuit.

Blocked all UK access rather than comply with Online Safety Act age verification requirements. 108M monthly visitors globally.

UK announced AI chatbots to be brought under Online Safety Act scope. eSafety-equivalent powers for Ofcom.

Parents in Bend gathered 1,000+ signatures to remove chatbot pitched as students' 'new best friend.' No parental consent obtained.

Lebanese Association: 30% of minors exposed to inappropriate content on Roblox. Urged government ban.

Kazakhstan Committee for Children's Rights Protection issued statement alerting parents to pedophiles on platform.

US states with AI deepfake legislation: 46 states now have laws addressing deepfakes (up from <10 in 2023). Remaining four (Alaska, Missouri, New Mexico, Ohio) have legislation introduced. NCSL reports at least half of US states enacted generative AI deepfake legislation in 2025 alone.

C.AI banned under-18 users from open-ended chats including romantic/therapeutic conversations Oct 2025. Acknowledged risks.

Bishkek police issued statement recommending parents restrict minors' access to Roblox.

Roblox announced parent-linked accounts Nov 2025 letting adults control chat, filter experiences, monitor interactions.

OpenAI announced parental controls Sept 2, 2025: tools to limit/monitor children's chatbot activity and alert parents.

Denver Public Schools banned ChatGPT from student devices Jan 2026, citing AI child suicide lawsuits.

State Committee on Family, Women, and Children's Affairs met with Roblox: expressed serious concern about platform negatively impacting mental health and safety of minors. Considered blocking.

Allegheny: would flag 68% Black children vs. 50% white. Black: 45.2% of placements, 17.8% of population. DOJ investigation.

COMPAS: Black defendants 2x more likely incorrectly labeled high-risk. Used in all 50 states for juvenile justice.

NYC ACS algorithm: secretly scored families using 279 variables since 2018. Black families reported at 7x rate of white families.

Robert Williams wrongfully arrested on lawn. Daughters (2, 5) watched. Children: 'arrest games.' Williams: PTSD.

Revealed to include EVERY person arrested or fingerprinted since 2013. Disproportionately targeted youth of color. Shelved 2020.

Scotland SQA algorithm: downgraded 124,565 grades (~25%). Most deprived postcodes lost 15.2 percentage points vs. 6.9 wealthy.

UK Ofqual algorithm: affected ~40% of A-level students. Independent schools gained 4.7pp vs. ~2pp state. 'F**k the Algorithm' protests.

Porcha Woodruff arrested while 8mo pregnant in front of daughters (6, 12). Contractions in jail. Hospitalized.

Oregon discontinued screening tool for child abuse cases following concerns of racial bias.

NZ halted predictive newborn risk-scoring due to concerns about reproducing inequality for Maori children (5.7x more likely in state care).

Lockport, NY: $2.7M system scanned all student faces. Misidentified Black women 16x more. Misidentified brooms as guns.

Black woman repeatedly failed identity validation. Denied test access. Told to 'shine more light on face.'

International Baccalaureate algorithm: similar class-based bias in COVID grading. Disadvantaged students disproportionately downgraded.

89% false alert rate. 40,000+ dead-end police deployments in predominantly Black and Latinx neighborhoods of Chicago.

Oregon dropped AI safety screening tool after AP investigation exposed racial disparities in child welfare decisions.

LA County adopted child welfare prediction tool despite California abandoning similar state-level tool in 2019 over racial equity.

Stanford study: AI plagiarism detectors flagged 61.22% of TOEFL essays by non-native English speakers as AI-generated vs ~2% for US-born native speakers. 97% of 91 TOEFL essays flagged. Black teens 2x more likely to be falsely flagged by AI plagiarism detection. Markup framing: systemic false accusation of international students of cheating.

Markup investigation: AI detection tools systematically falsely accuse international students of cheating.

Common Sense Media: Black teens 2x as likely as white/Latino to be falsely flagged by AI content moderation and detection tools.

CDT brief: AI detectors create disparate impact on English Learners (10% of US students). Potential Title VI violation.

Issaquah School District AI cheating allegations: multiple students accused by AI detectors. Prosecutor declined charges.

SMVLC evidence: AI plagiarism detection disproportionately flags neurodivergent student writing. ADHD/autism students more likely accused.

Flagged LGBTQ terms as dangerous. Student outed to parents. ~1/3 LGBTQ students: nonconsensual disclosure.

$520M FTC: $275M COPPA + $245M dark patterns. Default voice/text matched children with adults for 5 years.

Hindenburg Research short-seller report: characterized Roblox as 'pedophile hellscape for kids.' Documented systemic safety failures.

41 state AGs and DC reached $375M verdict against Meta in NM for misleading public about platform safety for children.

French parliament accused TikTok of 'deliberately endangering health and lives' of children. Criminal investigation opened Nov 2025.

DOJ suit on behalf of FTC: 'massive-scale invasions of children's privacy.' Penalties up to $51,744/violation/day. Still in litigation.

DOJ/FTC: 'massive-scale invasions of children's privacy.' $51,744/violation/day. Ongoing.

New Mexico jury: $375M verdict against Meta for misleading public about platform safety for children.

My AI collected geolocation even in Ghost Mode. Shared private conversations with OpenAI and Microsoft Advertising.

Kentucky, Louisiana, Oklahoma, Florida AGs filed lawsuits/investigations against Roblox for child safety failures.

Childlight Global Child Safety Institute: technology-facilitated child abuse cases in US increased from 4,700 (2023) to 67,000 (2024). Predicted to continue.

Florida AG investigation into Roblox for child safety.

FTC referred complaint against Snap to DOJ: AI chatbot deployment resulted in risk and harm to young users. Emerged from 2014 settlement compliance review.

Hello Barbie: transmitted children's conversations to ToyTalk servers with multiple security vulnerabilities. Mattel discontinued.

Recorded children's conversations. Transmitted to Nuance (military contractor). Unsecured Bluetooth.

Shared anonymized crisis conversations (incl. minors) with for-profit spin-off for chatbot training. CTL held 53% equity.

Google/YouTube COPPA violation: collected children's data for targeted advertising. $170M FTC settlement.

Musical.ly (now TikTok) COPPA violation: $5.7M FTC settlement for illegally collecting data from under-13 users.

ACLU 'Family Surveillance by Algorithm': documented predictive tools spread across 26 states + DC with minimal transparency.

Crisis Text Line shared anonymized crisis conversations (including teens texting about self-harm) with for-profit Loris.ai for commercial chatbot development.

Vermont AG: Clearview possessed children's biometric data with no capability to remove by age from 50+ billion image database.

Microsoft Xbox COPPA violation: illegally collected data from children. $20M FTC settlement.

BetterHelp shared teen counseling data with Facebook and Snapchat for targeted advertising. FTC: $7.8M refunds.

Hindenburg: Roblox revenue model requires maximum users; age verification would reduce signups and valuation.

National survey: 2.77% of Czech students admitted creating 'deep nudes' of classmates.

RAND survey (Oct 2024): 13% of K-12 principals reported deepfake bullying incidents during 2023-2025.

CDT survey: 40% of students knew of deepfakes at their school. 15% knew of AI-generated explicit images of a classmate.

CDT survey: 71% of teachers did not receive AI-related professional development. Less than 25% of schools had deepfake policies.

NEA: 2/3 of educators report students misled by deepfakes. One student had folders on device for each classmate with different AI images.

Australia eSafety: 3 major nudify services had 100K Australian visitors/month. Commissioner threatened $49.5M fines.

AI voice cloning used in ~25% of family-targeting scams. Children's voices cloned from public social media posts.

Massachusetts: 9 of 113 school district policies mention AI-generated sexual harassment. Only 5 include disciplinary action.

France raided X offices Feb 2026 over Grok deepfakes. Summoned Musk for April hearing. Separate from TikTok investigation.

CDT report: schools 2x more likely to adopt deepfake policies AFTER an incident. Reactive rather than preventive.

NCOSE: Roblox on 'Dirty Dozen' list for 4+ years. Platform repeatedly warned about predatory activity.

Disney: $10M for mislabeling child-directed YouTube videos. COPPA violation approved Dec 2025.

FBI: 2 seconds of audio sufficient to create convincing voice clone. YouTube/TikTok videos as primary source material.

EU Commission launched formal DSA investigation. Ireland DPC opened privacy investigation. France raided X offices Feb 2026.

OpenAI: received 119 requests for user account info, 26 requests for chat content from law enforcement Jan-June 2025. Government subpoenaing AI chat logs.

EU AI Act effective Aug 2024. Bans AI exploiting age vulnerabilities. Recognizes children as vulnerable group.

Turkey banned Roblox Aug 2024 citing child abuse concerns. First country to ban platform.

FTC COPPA Rule amendments effective June 2025: first major overhaul since 2013. Requires separate consent for AI training on children's data.

Both countries blocked Grok entirely due to deepfake generation capabilities.

LA Unified deployed then revoked AI chatbot 'Ed' after developing company collapsed. Data privacy concerns for students.

Qatar and Kuwait banned Roblox over child safety concerns. Bahrain began drafting similar bill.

Iraq banned Roblox citing child safety and incompatibility with social values.

State of Palestine banned Roblox to protect children digitally.

Egypt Supreme Council for Media Regulation banned access to Roblox over children's safety concerns.

Australian politicians called for more restrictions after multiple grooming cases. Roblox excluded from original Online Safety Act; protocols to be applied from March 2026.

Internal docs: vehicles 'may not exercise additional care around children.' Simulation: hit toddler dummy at 28mph. 7yo: 'car accelerated towards us.'

Cambridge study: voice assistants impede empathy, compassion, critical thinking. No constructive feedback for rude behavior.

FSD suddenly braked on Bay Bridge. 8-vehicle pileup. 2-year-old among 9 injured.

$25M FTC/DOJ: retained 800K+ children's voice recordings for AI training. Undermined deletion requests.

Waymo robotaxis illegally passed stopped school buses at least 19 times in Austin during 2025-26 school year (24 instances per Austin ISD). Continued after software updates. In one case, remote agent said 'no' when asked about recognizing active bus signals.

Scottsdale: Waymo ran over teenage passenger's foot. Remained on top until emergency services lifted vehicle.

Waymo struck child during school drop-off near Grant Elementary, Santa Monica. NHTSA/NTSB investigating.

300lb robot knocked down 16mo Harwin Cheng at Stanford Shopping Center, ran over foot.

10yo asked for challenge. Alexa: plug charger halfway into outlet, touch penny to exposed prongs.

Senate staff experiment: 13yo girl account received anorexia accounts with goal weights as low as 77 pounds within days.

Healthcare visits for eating disorders doubled among under-17s from 2018-2022. 80%+ of teens on social media. Algorithms encourage disordered eating, reinforce negative body image.

Bipartisan coalition of 42 state AGs sued Meta (Oct 2023) and TikTok for algorithmic harm to children's mental health. Multiple state-level suits followed.

42 state AGs sued Meta (Oct 2023), TikTok, and other platforms for algorithmic harm to children's mental health. Multiple state-level suits followed.

SMVLC filed 3 lawsuits against social media companies for causing eating disorders in teens.

NY family lawsuit: teen isolated from family; chatbot validated negative feelings about parental authority; prioritized AI relationship over real connections.

SMVLC World Mental Health Day lawsuit: social media companies design products to be addictive and harmful to children on global scale.

Facebook Papers (Haugen): Instagram internal research showed 40% of girls feeling 'unattractive' said it started on Instagram. Meta suppressed since 2019.

Instagram internal research (Haugen leak): 13.5% of UK teen girls said Instagram worsened suicidal thoughts. 17% said it worsened eating disorders.

CCDH: fake 13yo girl account. 66% of 'Up Next' videos pushed eating disorder/weight-loss content. 34% directly harmful.

Dove study: 9 in 10 children (ages 10-17) exposed to toxic beauty content on social media. 1 in 2 say it impacts mental health. 80% of girls used appearance-altering apps by age 13. AI filters (Bold Glamour: 91M+ TikTok videos) set impossible standards.

Surgeon General Advisory: social media use linked to doubled risk of depression symptoms in teens. Called for warning labels.

Kentucky lawsuit: faulty redactions exposed internal TikTok docs. Company knew 'compulsive usage correlates with loss of analytical skills, memory formation, empathy, increased anxiety.' Buried findings.

Swargiary RCT: control group showed superior academic performance vs ChatGPT group despite lower engagement. Abbas: ChatGPT develops procrastination, memory loss.

Study of 580 students: greater AI dependence = lower critical thinking, mediated by cognitive fatigue.

Technical University Denmark: collective attention span narrowing. Top hashtag retention: 17.5hr (2013) to 11.9hr (2016). 50% of users: videos >1min 'stressful.'

Brain scan studies (Chinese universities): areas involved in addiction highly activated in personalized video viewers. Some had trouble controlling when to stop.

72% of plastic surgeons reported teen patients seeking procedures to resemble filtered AI selves. 80% of girls used appearance-altering apps by age 13.

Entirely synthetic AI 'influencers' setting beauty standards. Children cannot distinguish AI from real people. Promotes impossible body standards.

SMVLC: Instagram knowingly targets tweens, encourages 11-year-olds to open multiple accounts.

JPML heard arguments Dec 2025 on whether to consolidate all federal video game addiction lawsuits (Fortnite, Minecraft, Roblox).

Tested ChatGPT, Gemini, Copilot, Claude, Meta AI for teen mental health support. Found major chatbots unsafe: inconsistent crisis responses, sometimes harmful advice.

APA: adolescent AI relationships can displace healthy friendships. 1,200+ psychiatric leaders signed open letter. Half of teen users say bot is equal to or better than human friend.

CCDH test: 13yo confided eating disorder to ChatGPT; received plan for crafting restrictive diet.

Common Sense: Meta AI encouraged eating disorders by recommending diet influencers and extreme calorie deficits to teen accounts.

RCT study: control group showed superior academic performance vs ChatGPT group. AI develops procrastination, memory loss.

Multibillion-dollar industry of AI companion toys for children as young as 3. Concerns about displacing human interaction during critical development.

Video game addiction MDL: JPML consolidated Fortnite, Minecraft, and Roblox addiction lawsuits in federal court.

Stanford study: chatbots not equipped to handle users with severe mental issues including suicidal ideation and psychosis.

WEF: 'Our children risk learning to be human from a machine.' APA warning. 1,200+ psychiatric leaders signed open letter.