The Home Office admitted the police national database's facial recognition search is more likely to wrongly include some demographic groups. National Physical Laboratory testing at a lower setting found false positive rates of 0.04% for white subjects, 4.0% for Asian and 5.5% for black subjects, and 9.9% for black women. The Home Office says a new algorithm will be tested early next year.
Nagoya prosecutors indicted a 34-year-old former elementary school teacher for possessing, in March, two sexual images generated with AI from pictures of two real girls, in Japan's first such case under its child pornography law. He told police they were nude images made on an AI site; he already faces trial in a teachers' voyeur-image-sharing case.
The New York Times sued Perplexity in federal court, alleging its AI search engine copied large chunks of Times content, sometimes entire articles, after repeated requests to stop, and falsely attributed hallucinated information to the paper. Perplexity's head of communication said publishers' suits against new tech have never worked, 'or we'd all be talking about this by telegraph.'
Asked for the addresses of 33 private individuals by name, xAI's Grok gave accurate current or former home addresses in 17 cases and refused only once, according to a Futurism investigation. The chatbot also volunteered addresses of possible relatives, along with current phone numbers and emails, though Grok's terms of service prohibit violating a person's privacy.
Police in Da Nang have arrested three South Korean men wanted on Interpol Red Notices, accused of running a Cambodia-based romance-scam network in which, police say, deepfakes let them pose as women to lure men, who were then trafficked, stripped of passports and forced to memorise scam scripts. Authorities estimate it scammed roughly US$1 million between August and December 2024.
A 64-year-old photographer in Pauillac was charged and remanded in custody on 26 November over the importation, possession and distribution of child sexual abuse images, Bordeaux prosecutors say, after allegedly using AI software to sexualise real photographs. A partial examination of seized devices already shows 'several thousand files' of such material, according to prosecutors.
Factcheckers Full Fact found hundreds of deepfake videos, built from real footage of doctors and influencers, urging menopausal women to buy supplements from US firm Wellness Nest, on TikTok, X, Facebook and YouTube. Wellness Nest says the videos are '100% unaffiliated' and it has 'never used AI-generated content', while TikTok removed one professor's 14 fakes six weeks after a complaint.
The ACLU of Northern California and partner firms sued landlord Equity Residential and vendor SmartRent in San Francisco Superior Court on behalf of three current and former tenants and the San Francisco Tenants Union, alleging forced 'smart home' systems violate tenants' constitutional privacy rights. The ACLU says such AI-enabled systems can track tenants' behaviour and when they are home.
Who did it: HYBRID AI Hallucination Cases database
A federal bankruptcy judge in San Diego fined a defence lawyer $950, ordered eight hours of training and announced a disciplinary referral, finding bad faith after an AI-drafted motion cited a nonexistent case and the lawyer misrepresented when a client signed a declaration. The lawyer first called the fake case a 'placeholder', then admitted the AI produced it.
