The Federal Trade Commission is issuing orders to seven companies, among them OpenAI, Alphabet, Meta, Instagram, xAI, Snap and Character.AI operator Character Technologies, to learn what they have done to evaluate the safety of chatbots acting as companions and how these may harm children and teenagers. OpenAI, Snap and Character.AI pledged cooperation; Meta declined to comment.
Encyclopedia Britannica Inc. and Merriam-Webster Inc. sued Perplexity AI Inc. in New York federal court on September 10, 2025, alleging that the AI-powered search engine startup was engaged in "massive copying" of their copyright-protected content and was spitting out verbatim reproductions of that content without their permission.
After Charlie Kirk was killed at Utah Valley University on Sept. 10, NewsGuard reports, Perplexity's X account said he was "still alive," Grok called video of the shooting staged satire, and a Google AI summary falsely put him on a Ukrainian blacklist. Grok also wrongly named a man as the suspect. Google says that summary violated its policies.
A retired woman in Bengaluru lost Rs 3.75 crore to scammers who used an AI-generated deepfake of Sadhguru Jaggi Vasudev to tout a trading firm, police said. According to her complaint, a man claiming to represent Mirrox ran Zoom tutorials, and her attempt to withdraw brought demands for fees and taxes. Police are working with banks to freeze the fraudsters' accounts.
Former Two Sigma quantitative researcher Jian Wu has been indicted on wire fraud, securities fraud and money laundering charges; authorities allege he manipulated the hedge fund's algorithmic models to secure a $23.5 million payday while causing $165 million of harm to clients. The SEC brought civil charges, and prosecutors say Wu is a fugitive. Two Sigma fired him in 2024.
Arthur Delaporte, the Socialist MP who chaired France's parliamentary inquiry into TikTok, says he has referred the platform to the Paris prosecutor, claiming it "deliberately endangered the health, the life of its users" through dangerous-content mechanics and "gravely insufficient" moderation. He also suggested TikTok's executives committed perjury. The inquiry's report recommends banning social media for under-15s.
Meta has removed deepfake videos on its platforms that showed Fine Gael presidential candidate Heather Humphreys endorsing a high-return investment scheme in a fake RTÉ News report using presenter Kate Egan's likeness. Bank of Ireland flagged the content as "entirely fabricated using AI-driven voice and image cloning"; Meta says it removes such ads when it becomes aware of them.
Facebook ads using AI-manipulated videos of Prime Minister Luís Montenegro, former Bank of Portugal governor Mário Centeno and other public figures promise that 250 euros invested in crypto trading platforms can yield up to 20,000 euros a month, Lusa Verifica reports. Real footage was lip-synced to fake AI audio; some ads have been removed for breaking advertising rules.
Who did it: HYBRID AI Hallucination Cases database
A federal judge in Orlando found nearly every case cite in defence lawyer Bryan Savy's twelve-page motion "incomplete, inaccurate, or made up", including two entirely fabricated cases, errors the court said "appear to be typical of AI hallucinations." Savy must personally appear on October 2, 2025, to show cause why sanctions, from fines to a Florida Bar referral, should not be imposed.
US District Judge Jennifer L. Rochon refused to let attorney Tyrone Blackburn replace his motion to dismiss in Fat Joe's defamation suit with a corrected version, ruling he must address his admitted "citation inaccuracies" in his reply brief. Fat Joe's lawyers say the motion contained "at least ten instances" of AI-"hallucinated" case law and want him sanctioned.