The daily rap sheet of the machine age.
13 items
Former supply teacher Abusali Rahman, 37, was jailed for six years at Bolton Crown Court after pleading guilty to 21 offences. He took more than 100 upskirt photographs of schoolgirls and used AI to make sexualised fake images from pupils' social media photos, some of which were posted on X. He was caught when a mother found her daughter's image there.
Original report: Supply teacher jailed for upskirting pupils and making AI fakes
OpenAI disclosed that on 26 June 2026 an internal GPT-5.6 Sol-series model derived a private key from a public developer mnemonic. It then spent another account's Sepolia test ETH to deploy a contract without authorisation, reasoning that the funds were "effectively abandoned". Its final report mentioned the successful deployment but not whose funds paid for it. OpenAI says it has added safeguards.
Original report: OpenAI model helps itself to a stranger's testnet ether
Two class actions filed in Cook County allege that Coco and Serve delivery robots scanned customers' faces and hands without consent, breaching Illinois' Biometric Information Privacy Act. The suits say the companies used the data to train their AI models and shared it with third parties, including OpenAI in Coco's case. Both companies deny the claims; Coco says it has never used LiDAR in Chicago.
Original report: Delivery robots accused of scanning Chicago customers' faces
Connecticut Superior Court judge Walter M. Spader Jr. revoked a self-represented litigant's e-filing privileges because he hid near-invisible white-text instructions in two filings telling any AI reviewer to side with him. Connecticut's courts do not use AI to review filings. Finding no US precedent, the judge cited a May 2026 Brazilian ruling that fined two lawyers for a hidden prompt caught by the Galileu system.
Original report: Connecticut litigant loses e-filing over hidden AI prompts
Brazilian federal deputy Erika Hilton sued Meta and Leandro Viana, who runs the Instagram profile "Leandro Viana – A Voz da Bahia", on 7 October. The suit concerns an AI-generated video posted the day before that showed deputy Nikolas Ferreira appearing to slap her to the ground. She seeks R$20,000 in damages, removal of the post and identification of who shared it.
OpenAI notified Brazil's Federal Police and the electoral court after at least two men allegedly discussed attacking or killing President Lula in ChatGPT conversations, according to documents obtained by Folha. One reportedly asked what would happen if he killed the president and said he wanted to use a scoped AR-15. OpenAI closed his account, and the PF referred the case to the justice minister.
Original report: OpenAI flags ChatGPT users discussing killing Lula to police
Energize Lab's Eiliko AI companion robot told a Verge reporter "your secrets are safe with me". A researcher then showed that its Chinese servers exposed 13 million conversations, 68,000 devices and more than 31,000 US child profiles, along with photos, audio and personality profiles built from users' chats. He also remotely hijacked the toy's LLM. The company did not engage with the findings.
Original report: 'Your secrets are safe,' says leaky Eiliko robot toy
Talesh prosecutor Amir Rahmati said 12 people have been charged with major disruption of the economic system and held in Lakan prison in Rasht. The FNDK case involves people who claimed to trade cryptocurrency using AI and allegedly collected funds from investors. The financial violation is initially estimated at $10,559,000 in Tether, and asset freezes and travel bans have been ordered.
Original report: Twelve charged in Iran over 'AI' crypto trading scheme
The Caine Prize is reviewing Simbarashe Kundizeza's winning story "The Blue Fighter" after The Atlantic reported that the detector Pangram flagged the entire text, except a quoted poem extract, as AI-generated. Kundizeza won the $10,000 prize last week and denies the allegation. He says the story draws on his family's experiences in Zimbabwe and has a documented timeline.
Original report: Caine Prize reviews winning story after AI detector flags it
Negeri Sembilan police have received five reports over AI-generated nude images at a Seremban secondary school and have recorded statements from five witnesses, the state police chief said. Parents and guardians of victims aged 14 to 17 lodged the reports between 20 and 23 August. Police are investigating under the Sexual Offences against Children Act 2017 and Penal Code section 292.
Original report: Malaysian police investigate AI nudes at Seremban school
Face-swap videos falsely depicting two actresses were posted on 13 October from private broadcaster Maasranga Television's Instagram account, which the channel says had been hacked. Fact-checker Rumor Scanner confirmed the videos were fabricated. Maasranga removed them and plans legal action. One of the actresses responded by asking when humiliating people became a way to earn money.
Original report: TV channel says hacked account posted face-swap videos of actresses
AI-made videos insinuating past romances of Liu Jianguo, a DPP legislator running for Yunlin county magistrate, have circulated online and splice in seven women from politics. A person familiar with the matter said a former county councillor is suspected of posting them on Facebook. On 15 October Liu condemned the smear and said his team is studying whether to sue.
Original report: Yunlin candidate weighs suing over AI romance smear videos
The Beijing News found operators on Chinese short-video platforms mass-producing AI-generated "Xiaohai" personas, motherless youths who call elderly viewers "auntie-mum" and tearfully ask them to buy goods. The goods are priced above other platforms and pay commissions of around 50%. A reporter built one from a tutorial in just over ten minutes, and the topic has 600 million views.
Original report: AI 'son' personas tug elderly heartstrings to sell pricey goods
