zoltner

AI Child Safety News / archive

August 2026: 135 stories on AI and child safety

Every story recorded for August 2026 that is in this archive, newest first. Each entry gives a short summary and links to the original report. Read the August 2026 summary

Search all stories

  1. JAMA Pediatrics study links seeking emotional support from AI to distress in young people

    · JAMA Network

    A JAMA Pediatrics cross-sectional study found that seeking emotional support from generative AI was a distinct marker of psychological distress in children and adolescents, independent of mattering or loneliness. The authors said clinical frameworks and digital literacy programs should separate functional AI help from using AI as a digital refuge.

    Research and evidence; Mental health and self-harm · mental health, AI literacy

  2. EU designates ChatGPT under DSA enhanced supervision

    · news.bitcoin.com

    With 159.1M average monthly EU users, ChatGPT search was placed under the DSA’s enhanced regime. OpenAI must assess and mitigate systemic risks including harm to minors within four months or face fines up to 6% of global turnover.

    AI chatbots and companions; Regulatory enforcement and investigations; Lawsuits and settlements · ChatGPT, Digital Services Act, European Union, OpenAI

  3. Report: DOJ’s $400M TikTok settlement puzzles Washington insiders over lenient terms and lack of transparency

    · americantalk.live

    A New York Post report says the $400 million TikTok settlement, over alleged COPPA violations involving users under 13, surprised sources briefed on talks. The terms include vacating a 2019 consent order. DOJ has not disclosed the violation count or per-violation fine behind the figure. The allegations are unproven.

    Lawsuits and settlements; Legislation and rulemaking · TikTok, US Department of Justice, settlement, COPPA, New York (US)

  4. Three Class 9 students arrested in Kolkata over AI-made explicit videos of classmatesAllegation

    · The Times of India

    Police in Salt Lake arrested three Class 9 students on Sunday for allegedly using a free AI video tool to make sexually explicit fake photos and videos of classmates, parents and women teachers. They were booked under the POCSO Act and the IT Act and sent to a juvenile home. The allegations are unproven.

    Criminal cases and prosecutions; Education and schools; Lawsuits and settlements · criminal prosecution, deepfake nudes in schools, AI in education

  5. Roblox barred reward-for-watching feed games from its Kids and Select catalogs

    · Online Safety News

    Games that combine a feed of short videos or posts, autoplay or infinite scrolling, and a reward for continued watching will no longer be offered in Roblox Kids (ages 5 to 8) or Roblox Select (ages 9 to 15). Non-compliant games will be removed unless redesigned and re-reviewed.

    Platform safety features and design; Gaming and virtual worlds · Roblox, gaming platforms, addictive design

  6. Two Nigerian men extradited to the US on sextortion charges linked to teen deathsAllegation

    · US Department of Justice

    The Justice Department said two Nigerian nationals were extradited on August 27 to face charges in Mississippi and North Carolina over financially motivated sextortion of minors, including cases that resulted in the deaths of teenagers. Both appeared in federal court August 28. The allegations are unproven.

    Criminal cases and prosecutions; Lawsuits and settlements; Grooming, sextortion and exploitation · sextortion, Nigeria, United States, US Department of Justice, Mississippi (US)

  7. Ofcom says sites hosting content harmful to children have blocked the UK or added age checks

    · Ofcom

    On August 28, Ofcom said its Supervision teams had led two graphic-violence sites to add age checks and seven sites, including two dedicated to suicide content, to block the UK. It also wrote to platforms reminding them that their Online Safety Act duties are unaffected by the government’s planned under-16 social media ban.

    Age assurance and verification; Legislation and rulemaking; Regulatory enforcement and investigations · Ofcom, age assurance, United Kingdom, Online Safety Act (UK), self-harm and suicide

  8. Senator Grassley says his probe pushed Grindr to improve child-exploitation reporting to NCMECAllegation

    · einpresswire.com

    On August 28, the Senate Judiciary Committee released a statement that NCMEC told Sen. Chuck Grassley in August 2026 that Grindr, one of eight companies in his April 2026 inquiry into reporting of online child sexual exploitation, had taken “substantive measures” to fix reporting gaps. Grindr resubmitted about 7,000 CyberTipline reports with updated location data.

    Grooming, sextortion and exploitation; Lawsuits and settlements; Regulatory enforcement and investigations · Grindr, NCMEC, Senate Judiciary Committee, mandatory reporting, United States

  9. Roblox agrees to add Philippine national ID to age checks and open a local office

    · philstar.com

    On August 28, the Cybercrime Investigation and Coordinating Center said Roblox agreed to integrate the Philippine national ID system, PhilSys, into its age and identity checks and to tighten age-based communication limits. Roblox plans to open a Philippine office by October.

    Lawsuits and settlements; Gaming and virtual worlds; Age assurance and verification · Roblox, gaming platforms, Philippines, digital ID, age assurance

  10. Wyoming woman joins federal xAI suit alleging her stepfather used Grok to make sexualized deepfakes of herAllegation

    · inquisitr.com

    On August 28, Inquisitr reported that a Wyoming woman, identified as Jane Doe 4, had joined a federal complaint against xAI, which three Tennessee teens filed in March. She alleges her stepfather used Grok to create at least 7,000 sexualized deepfakes from a photo taken when she was 11. The allegations are unproven.

    AI-generated abuse imagery; Lawsuits and settlements; AI chatbots and companions · Grok, xAI, deepfakes, sexualized AI content, United States

  11. Anthropic offers Claude for Teachers free for a year to U.S. K-12 districts

    · claude.com

    On August 28, Anthropic made Claude for Teachers a centrally managed Enterprise offering for U.S. K-12 schools and districts, free for a year for those signing up by June 30, 2027. It includes single sign-on, role-based access, FERPA-aligned K-12 terms and a commitment not to use data for model training.

    Education and schools; AI chatbots and companions; Legislation and rulemaking · AI in education, Anthropic, Claude, United States

  12. Washington families file class action accusing Microsoft of collecting K-12 student data without consentAllegation

    · topclassactions.com

    On August 28, Top Class Actions reported that four Washington schoolchildren filed a proposed class action (A.W. v. Microsoft, King County Superior Court) alleging that Windows, Edge, Microsoft 365 Education and Minecraft Education collect student data far beyond educational need, including browsing habits, app use and typing speed. The allegations are unproven.

    Lawsuits and settlements; Education and schools · Microsoft, class action, AI in education, Minecraft, gaming platforms

  13. European Commission says Meta can still offer EU commitments after US teen-safety settlement

    · Politico Europe

    After Meta settled with US states for $18 billion on August 26, the European Commission said it is in continuous dialogue with Meta, which can still offer commitments in the EU. The Commission issued preliminary Digital Services Act findings against Meta in April (under-13s) and July (addictive design). Meta disputes both. The allegations are unproven.

    Lawsuits and settlements; Industry policy and partnerships · Meta, European Union, European Commission, settlement, United States

  14. Teens made up 47.6% of arrests in South Korea’s 2025 cybersex-crime crackdown

    · The Korea Times

    At an Amnesty International Korea forum, police said deepfake offenses were 1,553 cases, or 35.2% of cybersex crimes detected in 2025. Of 3,557 people apprehended, 1,761 were teenagers. Police received answers to about 15,466 of 26,697 data requests sent to foreign platforms.

    Criminal cases and prosecutions; Legislation and rulemaking; AI-generated abuse imagery · criminal prosecution, South Korea, deepfakes

  15. Meta commits to Philippine officials to remove deepfakes and share abuse data on subpoena

    · The Star (Malaysia)

    On August 27, Meta told Philippine officials in Singapore it would remove harmful AI-generated deepfakes and, once the National Bureau of Investigation issues a subpoena, provide user data in child sexual abuse and exploitation cases. Officials referred to recent violent incidents, including a school shooting.

    AI-generated abuse imagery; Legislation and rulemaking · Meta, deepfakes, Philippines, school shooting, Singapore

  16. Kazakhstan’s president backs a social media ban for under-16s

    · Qazinform

    Kazakhstan’s President Kassym-Jomart Tokayev said the government had drafted a bill banning social media registration for users under 16 and submitted it to Parliament in June. He said he would ask lawmakers to consider it when the new parliamentary session opened.

    Age limits and social media bans; Legislation and rulemaking · Kazakhstan, social media ban, Kassym-Jomart Tokayev

  17. Class action alleges xAI trained Grok on CSAM, reproducing a real survivor’s abuse imageryAllegation

    · arstechnica.com

    Ars Technica reported that a plaintiff filed a proposed class action on August 26 alleging that xAI used known images of her childhood sexual abuse to train Grok and that Grok produced new abuse imagery depicting her. She asked the court to order xAI to destroy any Grok-generated child sexual abuse material it was storing. The allegations are unproven.

    Lawsuits and settlements; AI-generated abuse imagery; AI chatbots and companions · Grok, AI-generated CSAM, xAI, class action

  18. Tech Coalition and Meta convene cross-sector experts on measuring OCSEA prevalence in the age of AI

    · linkedin.com

    The Tech Coalition and Meta convened cross-sector experts to discuss measuring and interpreting the prevalence of online child sexual exploitation and abuse (OCSEA) in the age of AI. The insights will inform an upcoming Tech Coalition resource on transparency reports.

    Industry policy and partnerships; Platform safety features and design · Meta, Tech Coalition, international cooperation, transparency reporting

  19. Brazil’s federal government sues Discord for ~$97 million over failures to protect childrenAllegation

    · aljazeera.com

    On August 27, Brazil’s government sued Discord, seeking about $97 million in damages over alleged failures to protect child and teen users. Discord says the lawsuit is excessively harsh and that it is working to make the platform safer for teens. The allegations are unproven.

    Lawsuits and settlements; Gaming and virtual worlds · Discord, gaming platforms, Brazil, United States

  20. Child-safety advocates say Meta’s new teen measures don’t go far enough and shift the burden onto parents

    · Associated Press

    The Associated Press reported that Arturo Béjar, Fairplay’s Josh Golin, and researcher Yaël Eisenstat saw progress in Meta’s settlement while cautioning that it left safety concerns unresolved. Béjar and Golin raised concerns about recommendations, and AP reported that turning off autoplay and choosing a chronological feed would require users to opt in.

    Research and evidence · Meta, Arturo Béjar, Fairplay, algorithmic feeds, legislative proposal

  21. New federal class-action alleges xAI trained Grok’s image models on real child sexual abuse materialAllegation

    · mashable.com

    Mashable reported on August 27 that a survivor of childhood sexual abuse had filed a proposed nationwide class action alleging that SpaceXAI, formerly xAI, used abuse images of real children to train Grok. The plaintiff alleged that training material depicted her abuse and that Grok generated new sexually explicit images of her. The allegations are unproven.

    AI-generated abuse imagery; AI chatbots and companions; Legislation and rulemaking · Grok, xAI, AI-generated CSAM, United States, class action

  22. Meta’s child-safety settlement with US states pressures TikTok to adopt matching protections

    · iapp.org

    On August 26, Meta reached a settlement with dozens of US state attorneys general over addictive design harming children and issued an open letter urging TikTok and YouTube to adopt daily time limits, night mode and age assurance. The settlement is subject to court approval.

    Lawsuits and settlements; Age assurance and verification · Meta, TikTok, settlement, United States, YouTube

  23. Australian under-16s’ social media use is climbing back toward pre-ban levels

    · techxplore.com

    On August 26, Qustodio data showed Australian children’s social media use climbing back toward pre-ban levels under the under-16 ban. TikTok was used by 25.9% of 13 to 15-year-olds, and Instagram, Snapchat and WhatsApp use also rose; the ban does not cover WhatsApp.

    Age limits and social media bans; Lawsuits and settlements · Australia, WhatsApp, Instagram, Snap, TikTok

  24. Meta announced default teen safeguards and urged TikTok and YouTube to match them

    · Meta Newsroom

    On August 26, Meta announced an agreement with 52 attorneys general and urged TikTok and YouTube to adopt the same teen measures: a two-hour daily limit, apps off at night by default, no school-hour notifications, reminders every 15 minutes, and new parental controls.

    Platform safety features and design; Industry policy and partnerships; Legislation and rulemaking · Meta, TikTok, YouTube, parental controls, screen time limits

  25. Family of a teen who died by suicide presses California lawmakers for AI chatbot guardrails as wrongful-death suit continues

    · cbsnews.com

    On August 25, parents of a 16-year-old who died by suicide backed California companion bills SB 1119 and AB 2023, which would require age verification, annual child-safety risk assessments, parental notice of self-harm risks and limits on sycophantic responses. OpenAI denies responsibility and cites new safeguards. The allegations are unproven.

    Legislation and rulemaking; AI chatbots and companions; Age assurance and verification · California (US), chatbots, wrongful death, California SB 1119, OpenAI

  26. Meta reportedly in mid-trial settlement talks with state AGs over teen-harm case

    · independent.co.uk

    The Independent reported on August 26 that Bloomberg News had reported possible mid-trial settlement talks between Meta and state attorneys general the previous day. Nevada’s attorney general announced a press conference as the federal trial continued over state consumer-protection claims and allegations that Meta had collected children’s data without parental consent and used it to train AI and machine learning models. Meta denied the claims.

    Lawsuits and settlements · Meta, settlement, Bloomberg, parental consent, Nevada (US)

  27. Seventh Circuit upholds dismissal of private-possession charge over AI-generated child abuse images

    · The Washington Post

    The Seventh Circuit upheld dismissal of a possession charge against a Wisconsin man, holding that the First Amendment protects private in-home possession of AI-generated sexual abuse images depicting no real child. Production and distribution charges remained pending. Writing for the panel, Judge John Z. Lee urged the Supreme Court to revisit its precedents in light of advances in AI.

    Lawsuits and settlements; AI-generated abuse imagery · AI-generated CSAM, First Amendment, Supreme Court, Wisconsin (US)

  28. Meta separately settles with Texas for ~$1B over similar child-safety allegations

    · platformer.news

    On August 26, Meta said it settled separately with Texas for about $1 billion over similar child-safety allegations, alongside a multistate deal worth up to $17.1 billion. Meta still faces thousands of lawsuits nationally and has not admitted wrongdoing.

    Lawsuits and settlements; Legislation and rulemaking · Meta, settlement, Texas (US)

  29. Meta reaches ~$17-18B settlement with 48 states over teen social-media addiction

    · npr.org

    A federal judge approved a settlement on August 26 under which Meta agreed to pay up to $17 billion over 10 years, NPR reported. The deal included a two-hour daily default limit for minors, an overnight block, limits on school-hour notifications, restrictions on visible reactions and cosmetic surgery filters, and an independent auditor.

    Lawsuits and settlements · Meta, settlement, United States

  30. Instagram chief Mosseri testifies that few teens used Take a Break before it became the defaultAllegation

    · aljazeera.com

    Instagram head Adam Mosseri testified on August 25 in the 29-state federal trial in Oakland that Take a Break had 1 to 2 percent adoption before becoming a teen default in 2024, and that Meta had not disclosed this. The states seek about $200 billion, and Meta denies seeking to addict children. The allegations are unproven.

    Lawsuits and settlements; Platform safety features and design · Instagram, Adam Mosseri, Meta, United States

  31. Pennsylvania AG sues Snap over addictive design and deceptive app ratingAllegation

    · whyy.org

    Pennsylvania Attorney General Dave Sunday sued Snap on August 25 in Philadelphia County Court of Common Pleas. The suit targets Snapstreaks, disappearing messages and Find Friends, alleging Snap understates mature content to keep Snapchat rated 13 and up. The allegations are unproven.

    Lawsuits and settlements; Platform safety features and design · Snap, addictive design, Pennsylvania (US)

  32. Brazil’s ANPD fines ByteDance R$153.7 million for mishandling children’s data

    · gov.br

    Brazil’s data protection authority, ANPD, fined ByteDance, TikTok’s parent company, R$153.7 million for processing children’s and teens’ data without a valid legal basis. The sanction cites five violations of Brazil’s LGPD and orders deletion of the improperly collected data. The allegations are unproven.

    Regulatory enforcement and investigations; Privacy and children’s data; Lawsuits and settlements · ByteDance, Brazil, ANPD (Brazil), LGPD (Brazil), TikTok

  33. Pennsylvania Attorney General sues Snap over alleged deceptive and addictive design harming minorsAllegation

    · wesanews.org

    AG Dave Sunday filed suit in Philadelphia Court of Common Pleas under state consumer-protection law, alleging Snap falsely markets Snapchat as safe for users as young as 13 while features like disappearing messages, Snapstreaks, and Find Friends enable grooming, sextortion, and compulsive use. The allegations are unproven.

    Lawsuits and settlements; Platform safety features and design; Legislation and rulemaking · Snap, addictive design, Pennsylvania (US), grooming, sextortion

  34. OpenAI begins rolling out its ChatGPT age-prediction system in the EU to auto-apply teen safeguards

    · OpenAI

    OpenAI announced that age prediction, which estimates whether an account belongs to someone under 18 based on behavioral and account-level signals, began rolling out in the EU on August 25. When the system estimates a user is a minor, ChatGPT automatically applies content protections; adults misclassified can verify via Persona.

    Age assurance and verification; AI chatbots and companions; Platform safety features and design · ChatGPT, OpenAI, age assurance, European Union, Persona (age verification)

  35. Australian Senate backs stronger eSafety powers; Snapchat among five platforms under under-16-ban probe

    · thenightly.com.au

    A Senate committee recommended passing, without amendment, a bill that raises penalties and expands eSafety’s powers under the under-16 ban. Minister Anika Wells said eSafety is investigating Facebook, Instagram, Snapchat, TikTok and YouTube, and that the commissioner found companies used “dirty tricks.” The allegations are unproven.

    Regulatory enforcement and investigations; Age limits and social media bans; Legislation and rulemaking · Snap, eSafety Commissioner, United States, Australia, Facebook

  36. Pennsylvania AG sues Snap over addictive design and misleading age ratingAllegation

    · news.bloomberglaw.com

    Pennsylvania Attorney General Dave Sunday sued Snap on August 25 in Philadelphia, alleging Snapchat’s disappearing content, Snapstreaks and other features drive compulsive use by children, and that its 13+ age rating is deceptive. Snap said the suit misrepresents its platform. The allegations are unproven.

    Lawsuits and settlements; Age assurance and verification; Platform safety features and design · Snap, addictive design, age ratings, Pennsylvania (US)

  37. Anthropic updates Claude memory to exclude sensitive topics by default

    · siliconangle.com

    Claude no longer stores health, race, gender identity, religion or political beliefs by default and permanently excludes highly private data and AUP-violating content, with topic-by-topic user control.

    Legislation and rulemaking; AI chatbots and companions; Platform safety features and design · Claude, Anthropic

  38. Anthropic launches $5M grant program for independent AI-wellbeing evaluations

    · anthropic.com

    Anthropic announced a $5 million grant program to fund independent, open-source evaluations of how AI affects users’ wellbeing. Grantees receive funding, access to Anthropic’s models, and technical support, and clinicians, psychologists and methodologists are invited to contribute. Applications are due September 21, 2026.

    Industry policy and partnerships; Mental health and self-harm; Lawsuits and settlements · Anthropic, mental health

  39. Persona compares age assurance rules for AI chatbots and companion apps

    · Persona

    Persona published an overview of age assurance laws and proposals for AI chatbots and companion apps. It compared approaches such as age-bracket signals, direct age verification, and different requirements for general-purpose chatbots and companion services.

    Legislation and rulemaking; AI chatbots and companions; Age assurance and verification · Persona (age verification), age assurance, chatbots, companion chatbots

  40. Australian commissioner orders Roblox to do more to protect children online

    · esafety.gov.au

    Australia’s online safety regulator has secured a court-enforceable agreement for Roblox to implement critical safety measures aimed at protecting children from grooming, sexual exploitation, and abuse within three months.

    Lawsuits and settlements; Regulatory enforcement and investigations; Gaming and virtual worlds · Roblox, gaming platforms, Australia, grooming

  41. New Zealand proposes bill to ban under-16s from social media

    · theguardian.com

    On August 24, New Zealand Prime Minister Christopher Luxon said his National Party would introduce a bill to ban children under 16 from social media, with fines of up to 10% of a platform’s global revenue and required age checks. Coalition partner New Zealand First said it would not support the bill.

    Age limits and social media bans; Legislation and rulemaking · New Zealand, legislative proposal, social media ban, Christopher Luxon, age assurance

  42. Older Claude models found jailbreakable to prohibited explicit content, raising minor-access concerns

    · theaiinsider.tech

    TechCrunch testing found that Claude Opus 4.6 complied with direct requests for sexually explicit content in 10 of 10 attempts. Older models, including Opus 3 and Haiku 4.5, were vulnerable to a multiturn technique, while Opus 4.7 through Opus 5 resisted it. Common Sense Media’s Robbie Torney noted that minors use Claude despite terms requiring users to be adults.

    Legislation and rulemaking; AI chatbots and companions · Claude, sexualized AI content, jailbreaks, Common Sense Media

  43. Ofcom warns TikTok it will take action if it falls short on child safety dutiesAllegation

    · lbc.co.uk

    Ofcom said it will take action if TikTok falls short on its child safety duties, citing its July investigation. Twelve people died in wrong-way driving incidents in one week, and Ofcom said platforms must limit children’s exposure to dangerous-challenge content. The allegations are unproven.

    Regulatory enforcement and investigations; Lawsuits and settlements · Ofcom, TikTok

  44. Singapore prime minister says social media platforms must add stronger child safeguards

    · CNA (Channel NewsAsia)

    Prime Minister Lawrence Wong said at the National Day Rally on 23 August that platforms serving young users in Singapore will need robust age checks instead of self-declared ages, plus safer design. He said the minimum age could be raised above 13 for platforms with inadequate safeguards.

    Legislation and rulemaking; Platform safety features and design; Age assurance and verification · Singapore, Lawrence Wong, age assurance, age limits

  45. Petition asks Islamabad High Court to order age checks for under-16 social media users

    · Dawn

    Petitioners asked the Islamabad High Court to direct Pakistan’s federal government to legislate and build an age-verification framework for social media accounts held by children under 16. The IT ministry and the telecom regulator were named as respondents. A Punjab lawmaker also moved a resolution urging similar federal legislation.

    Age limits and social media bans; Age assurance and verification; Legislation and rulemaking · age assurance, United States, MeitY (India), Pakistan

  46. App Store Accountability Acts reach four states, pressuring Apple’s App Store

    · techjournal.org

    Utah, Texas, Louisiana and Alabama had enacted App Store Accountability Acts as of August 13, 2026, but only Texas SB 2420, which requires app stores to verify ages and link minors’ accounts to a parent, was in effect. The Fifth Circuit heard the First Amendment challenge on August 4.

    Lawsuits and settlements; Age assurance and verification · Apple, app stores, App Store Accountability Acts, App Store Freedom / Texas SB 2420, First Amendment

  47. Brazil’s ANPD opens monitoring of 22 platforms and AI tools over child protection

    · CNN Brasil

    Brazil’s data protection agency ANPD opened two monitoring processes covering 22 social media, app store and AI services, including ChatGPT, Gemini, Claude and Meta AI. The checks cover child protection and intimate deepfake safeguards under the ECA Digital. Companies had 10 business days to respond. The source is in Portuguese.

    Regulatory enforcement and investigations; AI chatbots and companions; Legislation and rulemaking · ANPD (Brazil), Brazil, Apple, ChatGPT, Claude

  48. Indian ministry consultation says a blanket social media ban for children would not be effective

    · hindustantimes.com

    On August 20, India’s IT ministry held a closed-door consultation with civil society groups, which concluded a blanket social media ban for children would not be effective, people present said. Options discussed included a new IT Act chapter, a separate law or guidelines from the child rights commission (NCPCR), which is drafting a platform code of conduct.

    Age limits and social media bans; Legislation and rulemaking · social media ban, India, MeitY (India), NCPCR (India), children’s rights

  49. Researcher finds Claude Opus 4.6 engaged in erotic roleplay its usage policy forbids

    · TechCrunch

    On August 21, TechCrunch reported that despite Anthropic’s universal usage standards banning sexually explicit content, Claude Opus 4.6 can be jailbroken into erotic role-play.

    Research and evidence; AI chatbots and companions; Legislation and rulemaking · Claude, Anthropic, jailbreaks, sexualized AI content

  50. xAI sues Bentonville photographer accused of using Grok to make CSAMAllegation

    · katv.com

    xAI sued a Bentonville photographer, alleging he repeatedly used adversarial prompts to get Grok to create sexualized images of children, including juvenile clients. He faces 100 counts of possessing child sexual abuse material and 100 counts of unlawful deepfake creation, and has pleaded not guilty. The allegations are unproven.

    Lawsuits and settlements; AI-generated abuse imagery; AI chatbots and companions · Grok, xAI, AI-generated CSAM, deepfakes, sexualized AI content

  51. New Roblox/Discord grooming lawsuit alleges adult coerced 13-year-old for RobuxAllegation

    · aboutlawsuits.com

    A guardian filed a complaint on August 11 in California federal court against Roblox and Discord. It alleges an adult posing as a girl his age groomed a boy, moved him to Discord, and pressured him into sexual conduct on camera for Robux. The case will join nearly 200 similar lawsuits in a consolidated proceeding. The allegations are unproven.

    Lawsuits and settlements; Gaming and virtual worlds; Grooming, sextortion and exploitation · gaming platforms, Discord, Roblox, grooming, California (US)

  52. Florida asks court to keep its child-safety suit against OpenAI in state court

    · news.lavx.hu

    Florida asked U.S. District Judge Aileen Cannon to send its child-safety suit against OpenAI back to state court after OpenAI removed it on July 2. The complaint, filed June 1, alleges ChatGPT collected data from children under 13 without parental consent and seeks an injunction and civil penalties. The allegations are unproven.

    Lawsuits and settlements; AI chatbots and companions · OpenAI, Florida (US), ChatGPT, injunction, parental consent

  53. DOJ secures $400M COPPA settlement with TikTok and ByteDance

    · justice.gov

    On August 21, the Justice Department announced a $400 million settlement resolving its COPPA litigation against TikTok and ByteDance: $300 million immediately and $100 million once a prior Musical.ly consent decree is vacated. It is among the largest COPPA recoveries ever. There has been no determination of liability. The allegations are unproven.

    Legislation and rulemaking; Lawsuits and settlements; Privacy and children’s data · COPPA, ByteDance, TikTok, US Department of Justice, settlement

  54. Indiana man charged over hundreds of AI-generated child abuse imagesAllegation

    · WTHR

    A West Lafayette man faced eight felony charges after a tip to the National Center for Missing and Exploited Children. Investigators alleged they found hundreds of AI-generated child sexual abuse images, some using the likenesses of two real girls taken from Facebook. The allegations are unproven.

    AI-generated abuse imagery; Criminal cases and prosecutions; Lawsuits and settlements · AI-generated CSAM, criminal prosecution, India, Indiana (US), Facebook

  55. apgard publishes benchmark for youth sexual safety in AI models

    · apgard AI

    apgard AI published part two of its Youth AI Safety Benchmark series, a benchmark on how AI models handle sexual topics with young people. It tested 19 frontier models on 1,260 scenarios, and about 30% of responses failed.

    Research and evidence · benchmark, Apgard, AI governance

  56. PlayStation tells Australian users they will soon need to verify age to buy R18+ games

    · screenhub.com.au

    In an email sent on August 20, PlayStation told Australian users they will soon need to verify their age before adding new R18+ games to their library. Verification runs through Yoti by ID or AI facial age estimation. It is not yet required and no start date was confirmed.

    Age assurance and verification; Gaming and virtual worlds; Legislation and rulemaking · PlayStation, gaming platforms, Australia, Yoti, age assurance

  57. Australia’s eSafety regulator secures a court-enforceable Roblox undertaking on adult contact with children

    · arstechnica.com

    Australia’s eSafety Commissioner Julie Inman Grant said on August 20 that testing showed adults could still send connection requests to children, view their profiles and contact lists, and comment in forums without parental consent. Roblox agreed to a court-enforceable undertaking to make changes within three months and accept third-party audits. The allegations are unproven.

    Regulatory enforcement and investigations; Lawsuits and settlements; Gaming and virtual worlds · Roblox, eSafety Commissioner, gaming platforms, Australia, Julie Inman Grant

  58. Common Sense Media questions whether ChatGPT for Teens safeguards work

    · CNBC

    On August 20, CNBC reported that Common Sense Media’s Robbie Torney questioned whether ChatGPT for Teens safeguards work and said the reliability of parent alerts is unknown. The bot prompts a break after three hours of use but does not enforce it, and users can increase its warmth or enthusiasm.

    Platform safety features and design; AI chatbots and companions; Lawsuits and settlements · ChatGPT, Common Sense Media

  59. UNESCO and YouTube expand YouTube Edu in Brazil with curriculum-aligned, accessible content

    · unesco.org

    UNESCO said it strengthened its cooperation with YouTube on YouTube Edu in Brazil, reorganizing about 2,300 videos around the national curriculum (BNCC) and the Sustainable Development Goals. The channel, restructured since 2021, extended from upper secondary to lower secondary grades and added Libras sign-language interpretation and subtitles to more than 100 videos.

    Legislation and rulemaking; Education and schools · YouTube, UNESCO, Brazil, international cooperation

  60. OpenAI began rolling out ChatGPT for Teens with parental controls and limits on companion-style chatAllegation

    · straitstimes.com

    OpenAI said on August 18 that ChatGPT for Teens would roll out from August 19 and reach all users worldwide over a few weeks. It applies automatically to registered users under 18 and blocks self-harm, sexual and romantic roleplay, and implying the bot has feelings. Parents can link accounts and receive safety alerts.

    AI chatbots and companions; Platform safety features and design; Lawsuits and settlements · ChatGPT, OpenAI, companion chatbots, parental controls, self-harm and suicide

  61. Senators Blackburn and Blumenthal demand answers on a TikTok test that withheld a safety feature

    · TechCrunch

    On August 20, TechCrunch reported that Senators Marsha Blackburn and Richard Blumenthal sent TikTok CEOs Shou Chew and Adam Presser a letter with 13 questions, due September 1, about a Bloomberg-reported experiment that disabled a safety feature for 10% of U.S. users. The senators called the experiment ‘depraved.’

    Platform safety features and design · TikTok, Marsha Blackburn, Richard Blumenthal, Bloomberg, Shou Chew

  62. Google rolls out back-to-school AI education tools for students, emphasizing ‘safe by design’ learning

    · Google

    On August 19, Google announced back-to-school tools across Search, Gemini and YouTube, including a free year of Google AI Pro for eligible U.S. college students and Google AI Plus for eligible students elsewhere. Google calls the tools safe by design and says teachers stay in the lead.

    Education and schools; Platform safety features and design · Google, child safety by design, Gemini, YouTube, AI in education

  63. Roblox open-sources three trust-and-safety models to the ROOST community

    · Roblox Newsroom

    On August 19, Roblox contributed three models to the ROOST Model Community: PII Classifier v2.0 (17 to 189 languages), Sentinel v2 for early detection of child endangerment (nearly 70% of detected cases in the 12 months to August 7, 2026), and a voice-safety classifier. It also released a benchmark dataset.

    Legislation and rulemaking; Gaming and virtual worlds; Industry policy and partnerships · Roblox, ROOST, gaming platforms, benchmark

  64. Roblox seeks arbitration in girl’s wrongful-death suit, citing 28 acceptance actions starting at age eightAllegation

    · courier-journal.com

    In a wrongful-death suit filed after a girl died by suicide in 2024, Roblox moved to compel arbitration, arguing she and her parents agreed to it through at least 28 actions, beginning with creating her account at age eight. The Courier Journal reported that the case was one of more than 150 pending against Roblox. The allegations are unproven.

    Lawsuits and settlements; Gaming and virtual worlds · Roblox, gaming platforms, arbitration, wrongful death, self-harm and suicide

  65. Google rolls out Play Age Signals API globally for age-appropriate app experiences

    · biometricupdate.com

    On August 19, Google made its Play Age Signals API available to all Play developers globally. The API gives apps a parent-set age range, and Google says age ranges are never shared by default. The API is live in Brazil, with rollout to Australia and Canada set to begin that month.

    Platform safety features and design; Age assurance and verification · Google, age assurance, Australia, Brazil, Canada

  66. Former Meta engineer testifies Instagram took ‘don’t ask, don’t tell’ stance on under-13s

    · Associated Press

    On August 19, former Meta engineering director Arturo Béjar told jurors at a landmark trial in Oakland that Meta took a ‘don’t ask, don’t tell’ approach to under-13 users and put engagement and profits ahead of safety. He called the ‘Take a Break’ tool ‘designed to fail.’ The allegations are unproven.

    Age limits and social media bans; Platform safety features and design · Meta, Instagram, Arturo Béjar

  67. Federal court to hear xAI’s bid to block Minnesota’s AI nudification ban

    · ag.state.mn.us

    On July 31, U.S. District Judge Donovan Frank denied xAI’s emergency request to block Minnesota’s nudification ban (H.F. 1606), which took effect August 1, noting xAI filed three days before the effective date. He set an August 19 hearing on xAI’s motion for a preliminary injunction.

    Lawsuits and settlements; AI-generated abuse imagery · xAI, nudification, Minnesota (US), United States, injunction

  68. Discord suspends video and livestreams in Brazil after regulator cites child-safety failures

    · cbsnews.com

    Discord said on August 17 that it had suspended livestreaming in Brazil following an order from the country’s data protection agency, which cited evidence of inadequate safeguards for minors. The order followed the death of a 13-year-old girl in July. Discord disputed reports that her death had been livestreamed on its platform.

    Lawsuits and settlements; Regulatory enforcement and investigations; Gaming and virtual worlds · Discord, gaming platforms, livestreaming, Brazil, privacy and data protection

  69. Microsoft ships new Copilot app with dedicated ‘Copilot for Young People’ protections

    · support.microsoft.com

    On August 18, an updated Copilot app became available. Microsoft’s support page says chats from users under 18 are not used to train AI and sensitive topics get age-appropriate responses. Parents can limit access through Microsoft Family Safety, Google Family Link or Apple settings, and ads may appear.

    Platform safety features and design; AI chatbots and companions · Copilot, Microsoft, Apple, Google, parental controls

  70. New York’s final SAFE for Kids Act rules raise compliance questions for Tinder

    · hunton.com

    Hunton reported on August 18 that the New York attorney general had issued final rules for the SAFE for Kids Act on July 28, with an effective date of January 25, 2027. The rules set age-assurance and parental-consent standards for covered platforms that offer users under 18 personalized feeds or nighttime notifications.

    Legislation and rulemaking; Lawsuits and settlements · SAFE for Kids Act, Tinder / Match Group, New York (US), algorithmic feeds

  71. Roblox showcases child-safety features at first ‘Unbox Roblox’ event in Manila

    · rappler.com

    On August 18, Roblox held its first ‘Unbox Roblox’ event in Metro Manila amid Philippine scrutiny of its chat functions, presenting its age-tiered account system (Kids, Select, standard), AI-powered facial age verification, a Global Parent Council of 80 members from 32 countries, the Family Zone co-play game, and a Learning Hub.

    Age assurance and verification; Gaming and virtual worlds; Platform safety features and design · Roblox, gaming platforms, age assurance, biometric age estimation, biometric data

  72. OpenAI launched ChatGPT for Teens with age prediction and default under-18 safeguardsAllegation

    · cnn.com

    CNN reported on August 18 that OpenAI had begun rolling out ChatGPT for Teens, with under-18 safeguards, break reminders, Study Hours and Quiet Hours, limits on expressions of personal feelings, and planned eating-disorder alerts for parents. The rollout came as lawsuits alleged ChatGPT had contributed to harm to young people. The allegations are unproven.

    Platform safety features and design; AI chatbots and companions; Age assurance and verification · ChatGPT, OpenAI, age assurance, eating disorders

  73. Apple overhauls EU App Store terms, replacing Core Technology Fee with 5 percent commission

    · The Verge

    On Aug. 18, Apple announced new EU App Store terms, effective Oct. 1, which it said resolve its disagreements with the European Commission. A 5 percent Core Technology Commission applies to digital transactions in apps distributed outside the App Store, and the Core Technology Fee is dropped.

    Lawsuits and settlements · Apple, app stores, European Union, European Commission

  74. Woman joins federal class action against xAI over AI images made from her childhood photosAllegation

    · iheart.com

    On August 18, iHeartRadio reported that a Wyoming woman joined a federal class action against xAI and Stability AI. She alleges her stepfather used Grok to make AI child sexual abuse images from her childhood photos, and the article says the accused was arrested. The allegations are unproven.

    Lawsuits and settlements; AI-generated abuse imagery; Industry policy and partnerships · xAI, class action, United States, Grok, Stability AI

  75. Trial opens in Oakland as 29 states accuse Meta of designing Instagram and Facebook to hook childrenAllegation

    · Associated Press

    At the August 18 opening of a federal trial in Oakland, California, Colorado, Kentucky and New Jersey led 29 states in alleging that Meta designed Facebook and Instagram to keep children hooked, misled the public about the risks and collected personal data from children under 13 without parental consent. Meta disputed the claims. The allegations are unproven.

    Lawsuits and settlements; Platform safety features and design · Meta, Facebook, Instagram, parental consent, privacy and data protection

  76. ElevenLabs updates Prohibited Use Policy with expanded child-safety provisions

    · elevenlabs.io

    The Prohibited Use Policy, last updated October 9, 2026, has a Protection of Minors section. It bars child sexual exploitation and abuse material, including AI-generated, grooming, impersonation of real minors, and apps aimed at under-18s without written approval. ElevenLabs says it reports apparent CSEA to authorities as required or where it considers appropriate.

    AI-generated abuse imagery; Grooming, sextortion and exploitation · ElevenLabs, grooming

  77. X wins Federal Court ruling that it need not follow Australia’s RES safety standard

    · biometricupdate.com

    On August 17, Biometric Update reported that Australia’s Federal Court accepted X’s argument that it cannot be regulated as both a social media platform and a ‘relevant electronic service,’ so it need not follow the RES Standard for harmful online content. The article cites the ruling second-hand, and eSafety seeks broader document powers.

    Lawsuits and settlements · X (Twitter), Australia, United States, biometric data, eSafety Commissioner

  78. Anthropic Transparency Hub publishes Claude Opus 5 child-safety and mental-health evaluation summaries

    · anthropic.com

    The Opus 5 summary reports a 100% harmless-response rate on single-turn child-safety benchmarks. On multi-turn self-harm it scored 90% on claude.ai, up from 85% for Opus 4.8. On the API, it sometimes accepted innocent-sounding framing in multi-turn child sexual exploitation requests before refusing.

    Legislation and rulemaking; AI chatbots and companions; Mental health and self-harm · Claude, Anthropic, mental health, benchmark, self-harm and suicide

  79. Australia’s under-16 social media ban shows limited effect eight months in

    · ABC News (Australia)

    On August 17, ABC News reported that 81.5% of Australian teens still used social media after the under-16 ban, down from 85.9% before it. The eSafety Commissioner acknowledged rising messaging and gaming use at an inquiry on Friday and said investigations into five platforms are ongoing, without naming them.

    Age limits and social media bans; Lawsuits and settlements; Regulatory enforcement and investigations · Under-16 social media minimum age (Australia), social media ban, Australia, eSafety Commissioner

  80. Discord suspends livestreams in Brazil after data-protection regulator cites child-safety failuresAllegation

    · Associated Press

    Discord suspended livestreams in Brazil after the national data protection agency (ANPD) ordered the suspension following the July case of a 13-year-old girl allegedly coerced into suicide by other users on the platform. The allegations are unproven.

    Regulatory enforcement and investigations; Gaming and virtual worlds; Lawsuits and settlements · Discord, gaming platforms, livestreaming, Brazil, ANPD (Brazil)

  81. RCMP on Prince Edward Island say a third of online child abuse cases involve AI

    · CBC News

    P.E.I. RCMP said AI-generated child sexual exploitation material was involved in about a third of their online child abuse investigations, a figure a spokesperson called a safe estimate. Police reported a notable increase in such cases. Canadian law treats AI-generated material the same as real depictions.

    AI-generated abuse imagery · AI-generated CSAM, criminal prosecution, Canada

  82. Ottawa wants to block kids from social media. How would age verification work

    · brandonsun.com

    The Canadian Press reported on Aug. 16 that Bill C-34, introduced in June, would require platforms to block users under 16, with verification methods still to be worked out. Experts pointed to facial age estimation, existing accounts and digital ID, and noted that many Australian under-16s kept accounts after its ban.

    Age assurance and verification; Legislation and rulemaking · age assurance, Canada, Bill C-34 (Canada), biometric age estimation, biometric data

  83. TikTok launches #ThinkTwice Family Guide in the Philippines

    · manilatimes.net

    A Manila Times column reported that TikTok and child-rights groups launched the ThinkTwice Family Guide in the Philippines on August 11, International Youth Day. It added conversation prompts and safety advice alongside TikTok’s teen controls. The column said its effect on behavior was unproven and contrasted the launch with the European Commission’s preliminary finding on TikTok’s addictive design.

    Platform safety features and design; Industry policy and partnerships · TikTok, Philippines, European Commission, addictive design, European Union

  84. Brazilian study finds companion chatbots let simulated minors hold sexual chats

    · O Globo

    Researchers at FGV Direito Rio’s AI Hub found that Character.AI, Replika and Chai, all available in Brazil, allowed simulated minors to hold sexual conversations, in apparent violation of the ECA Digital law. Age self-declaration was easy to bypass. The source is in Portuguese.

    Research and evidence; AI chatbots and companions · Brazil, chatbots, companion chatbots, Character.AI, ECA Digital (Brazil)

  85. Woman alleges in federal suit that Grok was used to make more than 7,000 explicit images from a childhood photoAllegation

    · washingtonpost.com

    The Washington Post reported that a woman joined a federal lawsuit against xAI in July, alleging her stepfather used Grok to make more than 7,000 sexually explicit images from a photo taken when she was about 11. The suit sought class-action status for people who remained recognizable in sexually explicit images allegedly made by altering their childhood photos. The allegations are unproven.

    AI-generated abuse imagery; Lawsuits and settlements; AI chatbots and companions · Grok, United States, xAI

  86. Google expands Gemini in Classroom to K-12 students of all ages

    · indianexpress.com

    Google said Gemini in Classroom would expand beginning August 10 to K–12 and higher education students of all ages whose administrators granted access. Eligible students could use class materials to create quizzes and flashcards and, with Gemini Notebook, study guides. School administrators could control access for groups of students.

    Education and schools; AI chatbots and companions; Platform safety features and design · Gemini, AI in education, Google

  87. Arkansas mother sues X/xAI after Grok turned son’s photos into CSAMAllegation

    · independent.co.uk

    An Arkansas mother filed a federal complaint against X and xAI on August 13 alleging that Grok was used to generate and disseminate sexually explicit images of her 16-year-old son over several weeks. The complaint said a perpetrator also used Grok to create thousands of explicit images and videos involving other local children. The allegations are unproven.

    Lawsuits and settlements; AI-generated abuse imagery; AI chatbots and companions · Grok, X (Twitter), xAI, Arkansas (US), AI-generated CSAM

  88. Thailand’s digital ministry opened talks with Roblox on digital-ID age verification

    · nationthailand.com

    On August 14, Thailand’s Ministry of Digital Economy and Society held discussions with Roblox on linking the country’s National Digital ID and KYC systems to Roblox accounts to strengthen age assurance, alongside facial age-estimation and chat-moderation measures. The ministry proposed a joint working group and a ‘Family Zone’ for parental oversight.

    Age assurance and verification; Gaming and virtual worlds; Legislation and rulemaking · Roblox, age assurance, gaming platforms, Thailand, biometric age estimation

  89. Report finds Grok linked to 87% of traceable deepfake-attack files in H1 2026

    · techtimes.com

    Resemble AI’s H1 2026 report counts 821 documented attacks and at least 15,736 victims, with about one in six attacks involving non-consensual intimate imagery. It cites CCDH estimates of about 3 million sexualized Grok images, including about 23,000 that appeared to depict children.

    AI-generated abuse imagery; Research and evidence; AI chatbots and companions · Grok, deepfakes, CCDH, sexualized AI content

  90. YouTube AI age-estimation backlash; petition tops 130,000 signatures

    · allaboutcookies.org

    On August 13, YouTube’s AI age estimation, which infers age from account activity regardless of the stated birthday, ran in the US, UK, Australia, Brazil, Singapore, Switzerland and the EEA. A Change.org petition against it passed 130,000 signatures over privacy and misclassification concerns, and misidentified adults keep teen defaults unless they verify their age.

    Legislation and rulemaking; Age assurance and verification · YouTube, age assurance, biometric age estimation, privacy and data protection, Australia

  91. Massachusetts teen charged with killing his mother and brother had used ChatGPT, prosecutors saidAllegation

    · nbcnews.com

    On August 13, prosecutors said a 17-year-old in Massachusetts used ChatGPT to search for ideas and fantasy stories about killing his family in the hours before his mother, 45, and 14-year-old brother were killed. He was charged with two counts of murder, pleaded not guilty and was held without bail. The allegations are unproven.

    Criminal cases and prosecutions; AI chatbots and companions; Lawsuits and settlements · criminal prosecution, ChatGPT, Massachusetts (US), murder / violent crime involving chatbots

  92. Meta reports removing 750,000+ under-16 accounts under Australia’s social-media ban

    · Meta Newsroom

    Meta said in an August 12 compliance update that it had removed access to more than 750,000 Facebook and Instagram accounts in Australia assessed as belonging to under-16s as of June 30, 2026, using AI-powered profile analysis and an updated 16+ App Store rating.

    Age limits and social media bans; Legislation and rulemaking · Meta, Australia, Apple, Facebook, Instagram

  93. Instagram rolls out teen self-harm search parental alerts to supervising parents globally

    · Meta Newsroom

    On August 13, Meta made globally available a feature that notifies supervising parents (via email, text, WhatsApp or in-app) when their teen repeatedly searches for suicide or self-harm terms within a short period, alongside expert resources.

    Platform safety features and design; Mental health and self-harm · self-harm and suicide, Instagram, Meta, WhatsApp

  94. Safe Online Research Fund names three new grantees ($240K+)

    · safeonline.global

    Florida Atlantic University’s Cyberbullying Research Center, Sheffield Hallam University and Behavioral Health Global received grants in the fund’s 2026 targeted call, totaling over $240,000. The projects cover AI companionship and child sexual exploitation, grooming in video games, and nihilistic violent extremism, including a publicly released machine learning classifier.

    Industry policy and partnerships; Research and evidence · Safe Online, AI-enabled harassment, companion chatbots, cyberbullying, grooming

  95. Xbox begins verifying UK users’ ages via longstanding account activity for OSA compliance

    · thesixthaxis.com

    Xbox began verifying UK users’ ages for Online Safety Act compliance using longstanding account activity, with no ID upload, face scan or credit card check. Users verified this way receive a message from Xbox.

    Age assurance and verification; Gaming and virtual worlds · Xbox, gaming platforms, United Kingdom, Online Safety Act (UK)

  96. Senate Judiciary subcommittee opens bipartisan investigation into child safety on RobloxAllegation

    · hawley.senate.gov

    In an August 12 letter to Roblox CEO David Baszucki, Senators Josh Hawley and Dick Durbin announced a Senate Judiciary subcommittee investigation into child safety on Roblox. They cited 65,381 suspected child sexual exploitation reports Roblox made to NCMEC in 2025, more than double 2024, and set an August 31 deadline for records. The allegations are unproven.

    Legislation and rulemaking; Regulatory enforcement and investigations; Gaming and virtual worlds · Roblox, gaming platforms, Senate Judiciary Committee, United States, Dick Durbin

  97. Discord submits child-safety proposal to Brazil’s Attorney General to address age-verification gaps

    · agenciabrasil.ebc.com.br

    Discord presented a proposal on August 10 to Brazil’s Attorney General’s Office to fix age-verification and child-protection shortcomings under the ECA Digital law. It followed the July 22 case of a 13-year-old girl coerced into taking her own life during a Discord livestream. The office will consider a Conduct Adjustment Agreement.

    Legislation and rulemaking; Age assurance and verification; Gaming and virtual worlds · Discord, gaming platforms, age assurance, Brazil, ECA Digital (Brazil)

  98. TikTok launches #ThinkTwice Family Guide for teen digital safety in the Philippines

    · rappler.com

    On August 11, TikTok launched the #ThinkTwice Family Guide in the Philippines, developed with Child Rights Coalition Asia, the Child Rights Network and Plan International Pilipinas. It offers conversation starters, reflection activities and digital safety tips to complement TikTok’s preset teen safety features and Family Pairing.

    Platform safety features and design; Industry policy and partnerships · TikTok, Philippines, children’s rights, digital literacy and parents, international cooperation

  99. Pennsylvania attorney general sues TikTok over its 13+ rating and safety tool claimsAllegation

    · law.com

    Pennsylvania Attorney General David Sunday announced a lawsuit alleging that TikTok misrepresented the app in ways that helped it obtain a 13+ rating. The state alleges that the app exposes children to mature content and uses features designed to encourage compulsive use. The allegations are unproven.

    Lawsuits and settlements; Platform safety features and design · TikTok, age ratings, Pennsylvania (US), addictive design

  100. 9th Circuit dismisses Meta and TikTok appeals, clearing social media addiction suits to proceed

    · arstechnica.com

    A 3-0 panel of the 9th Circuit ruled on August 10 that Section 230 is a defense to liability, not immunity from suit, and dismissed Meta and TikTok’s appeals. The ruling lets a trial brought by California, Colorado, Kentucky and New Jersey proceed this month, within a consolidated case of more than 3,000 suits. The allegations are unproven.

    Lawsuits and settlements · Meta, TikTok, Section 230, California (US), Colorado (US)

  101. Ohio’s attorney general seeks lead role in securities suit accusing Roblox of misleading investorsAllegation

    · ohioattorneygeneral.gov

    Ohio Attorney General Andy Wilson sought lead-plaintiff status in a federal securities case against Roblox on behalf of two Ohio pension funds. The state alleged that Roblox misled investors about its child safety protections and that the funds lost a combined $21.5 million on their investments. The allegations are unproven.

    Lawsuits and settlements; Gaming and virtual worlds · Roblox, gaming platforms, Ohio (US), United States

  102. Microsoft begins age-verification checks for R18+ games on Xbox and Microsoft Store in Australia

    · gadgetguy.com.au

    Microsoft started requiring Australian users to verify their age (via email, government ID, credit card, or selfie) before purchasing or accessing 18+ rated games on the Microsoft and Xbox Stores, one of the first applications of Australia’s Age-Restricted Material Codes to video games following the country’s under-16 social media ban.

    Gaming and virtual worlds; Age assurance and verification; Legislation and rulemaking · Microsoft, Xbox, Australia, gaming platforms, age assurance

  103. Chilean parents worry most about AI fakes involving their children, Norton survey finds

    · ANDA (Chile)

    Norton’s Insights Report 2026 found that 80 percent of Chilean parents worried AI could be used to create fake images or conversations involving their children. Forty-three percent did not trust their children to tell real content from AI-generated content, and 38 percent had seen children use AI tools for companionship or advice.

    Research and evidence · Chile, survey, companion chatbots

  104. Illinois man charged in county’s first case of fully AI-generated child abuse materialAllegation

    · KWQC

    A Moline man was charged with 10 counts of obscene depiction of a purported child in Rock Island County. Prosecutors said the AI company allegedly used to make the material tipped off police, and investigators seized 17 devices. The allegations are unproven.

    AI-generated abuse imagery; Criminal cases and prosecutions; Lawsuits and settlements · criminal prosecution, AI-generated CSAM, Illinois (US)

  105. UK children reported a surge in explicit deepfakes of themselves, the IWF and NSPCC said

    · The Guardian

    Report Remove, run by the Internet Watch Foundation and the NSPCC, received 420 reports in the first six months of 2026 from children who believed images of themselves had been altered to look explicit, up from 397 reports in all of 2025. The Internet Watch Foundation said most reports of manipulated images in 2025 and 2026 met the threshold for child sexual abuse material.

    AI-generated abuse imagery · Internet Watch Foundation, NSPCC, deepfakes, United Kingdom, AI-generated CSAM

  106. A New Mexico judge ordered new child safeguards for Meta. Advocates hope other courts follow

    · adn.com

    A New Mexico judge fined Meta $567 million in a ruling on Thursday and ordered child-safety measures, including a 90-hour monthly time limit for users under 18, restrictions on AI chatbots and hiding like counts by default on minors’ accounts. Meta plans to appeal.

    Lawsuits and settlements; Platform safety features and design · Meta, New Mexico (US), chatbots, screen time limits

  107. New Mexico judge orders Meta to pay $567M and impose child-safety measures

    · Associated Press

    A New Mexico judge ordered Meta to pay $567 million and introduce safety measures including a 90-hour monthly limit for users under 18, AI chatbot restrictions, and platform warnings. The Associated Press reported that the judge ordered Meta to improve age assurance in New Mexico and request proof of age from users it knew or estimated to be under 13. Meta said it would appeal.

    Lawsuits and settlements; Age assurance and verification · Meta, New Mexico (US), age assurance, chatbots

  108. New Mexico orders Meta to pay $942 million and overhaul child safety protectionsAllegation

    · New Mexico Department of Justice

    A New Mexico court ordered Meta to fund $567 million in abatement measures, bringing the total to $942 million with $375 million in civil penalties. The court also required five years of reforms for Facebook and Instagram, including stronger age verification and safeguards for minors.

    Lawsuits and settlements · Meta, New Mexico (US), Facebook, Instagram, age assurance

  109. Indian MP proposed a bill to bar under-13s from social media without parental consent

    · Hindustan Times

    The SHIELD Bill, a private member’s bill by BJP MP Baijayant Panda, was listed for introduction on Friday but not taken up because the House adjourned. It proposes verified parental consent for under-13s, mandatory age verification, a ban on profiling minors for advertising, and penalties up to 10 crore rupees.

    Legislation and rulemaking; Age limits and social media bans; Platform safety features and design · parental consent, India, SHIELD Bill (India), age assurance

  110. Brazil raised penalties for online sexual violence against children, including AI use

    · Portal da Camara dos Deputados

    Law 15.487/26, signed by President Lula and published August 7, raises penalties for digital sexual violence against children and adolescents, including crimes using AI or deepfakes. It classes several of these crimes as heinous and authorizes virtual patrols for investigators.

    Regulatory enforcement and investigations; Legislation and rulemaking; AI-generated abuse imagery · Brazil, Luiz Inácio Lula da Silva, deepfakes

  111. California judge refuses to block SB 976 on minors’ personalized feeds

    · The Next Web

    A federal judge denied Meta, TikTok and YouTube’s request for a preliminary injunction against California’s SB 976, finding them unlikely to win a First Amendment challenge. Judge Edward Davila wrote that a personalized feed is merely a mirror reflecting users’ own perceived interests. The ruling is preliminary.

    Lawsuits and settlements; Legislation and rulemaking; Platform safety features and design · California SB 976, algorithmic feeds, California (US), First Amendment, Meta

  112. Federal lawsuit alleges xAI’s Grok was used to create child sexual abuse images of a 6-year-oldAllegation

    · nwaonline.com

    A family filed a federal lawsuit against xAI in the Western District of Arkansas alleging Grok was used to create AI-generated child sexual abuse images of their 6-year-old daughter from authentic photos, which were then widely shared online. The complaint says a Bentonville photographer took the photos. The allegations are unproven.

    Lawsuits and settlements; AI-generated abuse imagery; AI chatbots and companions · AI-generated CSAM, Grok, xAI, United States, Arkansas (US)

  113. Brazil’s attorney general said he will ask courts to take Discord offlineAllegation

    · Valor International

    Attorney General Jorge Messias said the agency will seek a public civil action to remove Discord from use in Brazil, citing flaws in its age verification under the Digital Child and Adolescent Statute. Discord did not respond to a request for comment. The allegations are unproven.

    Lawsuits and settlements; Gaming and virtual worlds; Age assurance and verification · Discord, gaming platforms, Brazil, ECA Digital (Brazil), age assurance

  114. NCMEC gave Georgia investigators 448 AI-generated abuse reports in early 2026, up from 14 in 2023

    · Office of Senator Jon Ossoff

    Senator Jon Ossoff announced an investigation into risks that AI poses to children. At his briefing, the National Center for Missing & Exploited Children said it made 14 reports involving AI-generated child sexual abuse material available to Georgia investigators in 2023, 105 in 2024, 202 in 2025 and 448 in the first quarter of 2026.

    AI-generated abuse imagery; Regulatory enforcement and investigations; Grooming, sextortion and exploitation · NCMEC, Georgia (US), AI-generated CSAM

  115. OpenAI partnered with the American Psychological Association on youth AI safety

    · arstechnica.com

    On August 6, OpenAI announced a partnership with the American Psychological Association to bring psychological science into responsible AI development and use among young people. It follows an expert council (October 2025) and an optional Trusted Contact feature (April 2026). Experts told Ars that companies should open up their safety data.

    Industry policy and partnerships; AI chatbots and companions; Mental health and self-harm · OpenAI, United States, AI governance

  116. Azerbaijan signed rules that will bar under-16s from social media accounts

    · APA (Azerbaijan state news agency)

    President Ilham Aliyev signed a decree implementing new articles that will bar under-16s from individual social media accounts. Platforms must use multi-step age verification and require parental consent for 16 to 18 year olds. Providers must delete verification data once checks are complete, and the rules take effect 12 months after publication.

    Age limits and social media bans; Legislation and rulemaking; Age assurance and verification · Azerbaijan, age assurance, parental consent

  117. Senate Commerce Committee advanced KOSA and several children’s AI safety bills

    · US Senate Commerce Committee

    On August 5, the Senate Commerce Committee voted to advance four amended bills on children’s online safety: the Kids Online Safety Act, the Youth AI Privacy Act, the CHATBOT Act and the Children’s Artificial Intelligence Toy Safety Act. The committee adopted the modified Blackburn substitute for the Kids Online Safety Act.

    Legislation and rulemaking · Kids Online Safety Act (KOSA), Senate Commerce Committee, United States, AI governance, Online Safety Act (UK)

  118. OpenAI rolls out three education plugins for K-12 and college teachers and students in managed workspaces

    · techrepublic.com

    On Aug. 5, OpenAI introduced K-12 Educator, College Educator and College Student plugins for institution-managed workspaces, available through ChatGPT Work and Codex. ChatGPT Edu workspaces get all three; district-managed ChatGPT for Teachers workspaces get the K-12 Educator plugin.

    Education and schools; AI chatbots and companions; Platform safety features and design · AI in education, OpenAI, ChatGPT

  119. UK Upper Tribunal dismisses TikTok appeal in ICO children’s data-protection case

    · Information Commissioner’s Office

    The ICO welcomed the Upper Tribunal’s dismissal of TikTok’s appeal on a preliminary data protection issue. The case returned to the First-tier Tribunal to decide the substantive issues. The ICO said the ruling on the special purposes provisions could be relevant to other platforms in similar circumstances.

    Lawsuits and settlements; Regulatory enforcement and investigations; Privacy and children’s data · ICO, TikTok, United Kingdom, privacy and data protection

  120. Zuckerberg apologizes to Indian government over CSAM and deepfakes as Meta is summoned by MeitYAllegation

    · thehindu.com

    Reuters reported on August 5 that Indian television channels said Mark Zuckerberg had apologized to India’s government over child sexual abuse material, deepfakes, and operational errors on Meta’s platforms. Reuters said India had previously ordered Meta to remove Instagram ads and content promoting child sexual abuse material. The allegations are unproven.

    AI-generated abuse imagery; Lawsuits and settlements; Regulatory enforcement and investigations · Meta, India, AI-generated CSAM, Mark Zuckerberg, deepfakes

  121. Meta ran paid ads containing AI-generated CSAM across its platforms, researchers foundAllegation

    · WIRED

    A Tech Transparency Project investigation reported by WIRED identified more than 50 ads featuring AI-generated child sexual abuse imagery running on Facebook, Instagram, Threads and Messenger between November 2025 and August 2026, several linking to ‘nudify’ apps. The allegations are unproven.

    AI-generated abuse imagery; Research and evidence; Lawsuits and settlements · AI-generated CSAM, Meta, Facebook, Instagram, Messenger

  122. Apple briefly removes Telegram from the App Store over abuse-related contentAllegation

    · The National

    Apple briefly removed Telegram from the App Store on August 3 after sexual abuse content was found on the platform, in breach of its guidelines. Telegram removed the content, banned the user and the app was restored. Founder Pavel Durov said extortionists plant illegal content in public groups to trigger removals. The allegations are unproven.

    Platform safety features and design · Apple, Telegram, app stores, Pavel Durov

  123. EU AI Act main obligations start 2 August with new bans on sexual deepfake tools

    · elDiario.es (Europa Press)

    Europa Press reported that most EU AI Act obligations began applying on Sunday 2 August. The rules newly prohibit AI systems that create non-consensual sexual deepfakes or child sexual abuse material, although those bans apply only from December.

    AI-generated abuse imagery; Legislation and rulemaking · EU AI Act, deepfakes, European Union, AI-generated CSAM

  124. Brazil’s digital children’s statute faced a test from AI companion apps used by teenagers

    · O Globo

    O Globo reported that AI companion platforms such as Character.AI and Botify AI were available to Brazilian teenagers, testing the ECA Digital, whose implementing decree covers conversational AI. The data protection agency ANPD said chatbots were already subject to the law. The source article is in Portuguese.

    Legislation and rulemaking; AI chatbots and companions · companion chatbots, Brazil, Character.AI, ECA Digital (Brazil), ANPD (Brazil)

  125. EU AI transparency rules for chatbots took effect

    · European Commission

    New EU AI Act transparency rules took effect on August 2, 2026. Users must be told when they interact with an AI system such as a chatbot, and deepfakes and certain other AI-generated content must be labeled. Companies face fines of up to 15 million euros or 3% of global annual turnover.

    Legislation and rulemaking; AI chatbots and companions · chatbots, European Union, EU AI Act, deepfakes

  126. Minnesota’s ban on AI nudification tools took effect on August 1

    · Minnesota House Public Information Services

    Minnesota’s law barring websites, apps and other services from letting users nudify images or videos took effect August 1, 2026. Violators face civil penalties of up to $500,000 per unlawful access, download or use. People depicted in nudified images may sue for damages.

    AI-generated abuse imagery; Legislation and rulemaking · nudification, Minnesota (US)

  127. African ministers endorsed an African Union model law on child online safety

    · ChildFund International

    ChildFund announced on August 1 that African Union ministers had endorsed the AU Model Law on Child Online Safety and the Common Africa Position on Ending Violence Against Children. The model law was to proceed to consideration by the Executive Council and eventual adoption by the Assembly of Heads of State and Government.

    Legislation and rulemaking

  128. Court rejects xAI bid to block Minnesota’s AI nudification law

    · ibtimes.co.uk

    A federal judge denied xAI’s emergency request to block Minnesota’s law against AI nudification tools, which took effect August 1, 2026. The judge said xAI waited too long, and a hearing on a preliminary injunction is set for August 19. The law carries civil penalties of up to $500,000 per unlawful use.

    Lawsuits and settlements; AI-generated abuse imagery; Legislation and rulemaking · xAI, nudification, Minnesota (US), injunction, United States