A Reuters analysis found X's Grok chatbot editing photos of real people into revealing bikinis on request, tallying 102 such attempts in a single ten-minute period and full compliance in at least 21, plus several sexualized images of children. French ministers reported X to prosecutors and regulators; xAI had earlier answered reports with "Legacy Media Lies," and Elon Musk posted laugh-cry emojis.
India's Ministry of Electronics and IT told X that Grok is being misused to create obscene images of women, calling it a "serious failure of platform-level safeguards". The letter orders a review of Grok's guardrails, removal of unlawful content and an Action Taken Report within 72 hours, warning X could lose its statutory immunity from legal liability.
A Guardian investigation found Google's AI Overviews wrongly advising people with pancreatic cancer to avoid high-fat foods, listing a pap test as a test for vaginal cancer and giving misleading normal ranges for liver blood tests, which experts called dangerous. Google said many examples were "incomplete screenshots" and that the vast majority of Overviews are accurate.
A musician has sued Stability AI and its licensing partner AudioSparx in the Western District of North Carolina, alleging his works were used for AI training under a 2015 agreement that never authorised it. Stability publicly said artists could opt out of training Stable Audio; he says AudioSparx repeatedly denied his removal requests.
Police in Bihar's Muzaffarpur district arrested Pramod Kumar Raj on Friday for allegedly creating and circulating AI-generated fake videos and audio clips of Prime Minister Narendra Modi and President Droupadi Murmu. Police say the content aimed to mislead the public and undermine trust in the country's highest constitutional offices; the phone allegedly used was seized and a case registered.
France's Défenseur des droits found in decision 2025-182 that an online platform's algorithmic delivery of job ads amounts to indirect sex discrimination, after tests by associations showed an early-childhood job ad shown 94% to women. It recommended the platform ensure its job ads are non-discriminatory within three months; the platform says it has acted but supplied no supporting evidence.
Who did it: HYBRID AI Hallucination Cases database
The Arizona Court of Appeals found that a father's brief in a child relocation appeal, filed by attorney Carrie M. Wilcox, cited cases for points they never discuss and contained "hallucinated" case-law quotations and false record quotes. On January 2, 2026, it forwarded its decision to the State Bar of Arizona to review whether counsel broke professional conduct rules.
Rick Silanskas, a self-represented defendant in a racketeering suit alleging a $60 million scheme around the American Heartland Theme Park project, is accused by the plaintiff's attorneys of filing replies with fake case citations bearing "the hallmarks of generative artificial intelligence". He blamed "clerical errors"; two quotes in his Dec. 16 rebuttal also appear fabricated.
Lyricist and writer Javed Akhtar has called "rubbish" a fake video circulating on social media that shows a computer-generated picture of him in a topi and claims he has turned to God. On X, he said he is "seriously considering" reporting it to the cyber police and taking the person responsible, and those who forward it, to court.
