Deepfake ads using the face and voice of German doctor Eckart von Hirschhausen to sell slimming pills, potency drugs and painkillers are the subject of an ARD documentary airing 4 May. A study for it found over 750,000 people in Germany and Austria saw such ads on Meta in four weeks; only about 30 percent were removed.
Cape Breton fiddler Ashley MacIsaac is suing Google for $1.5 million, claiming its AI overview falsely said he had been convicted of offences including sexual assault and internet luring involving a child, and was on the sex offender registry. A First Nation confronted him with the summary and cancelled a concert, later apologising. The claims are untested in court.
Answering Sony Music's amended copyright complaint in New York federal court on April 29, Udio admitted it "obtained audio data from YouTube for use as training data" and acquired some training data using YT-DLP. It denies infringement, calls any copying fair use and accuses the labels of copyright misuse.
An attacker's Morse-code post on X was reportedly decoded by Grok into a plain instruction tagging Bankrbot, which then sent 3 billion DRB tokens, worth roughly $155,000 to $200,000, from a Grok-assigned wallet. A Bankr developer said 80% had been returned and that an earlier hardcoded block on Grok's replies was dropped in a rewrite.
An audit by Africa Uncensored, Lighthouse Reports and the Guardian found the Social Health Authority's means-testing algorithm, which President William Ruto has described as AI-powered, systematically overestimates poor Kenyans' incomes and overcharges them, while undercharging the wealthiest. A pre-launch IDinsight report shared with the government had called the system "inequitable".
A campaign ad from a group aligned with President Donald Trump uses AI-generated video showing Rep. Thomas Massie on a dinner date and holding hands with Reps. Ilhan Omar and Alexandria Ocasio-Cortez, as a narrator says he was "caught in a throuple". Massie called it defamatory, Omar demanded its removal and Marjorie Taylor Greene said it violates the Take It Down Act.
Taoyuan prosecutors have indicted a 51-year-old man surnamed Wang under the food safety law, saying he posted unverified AI-faked news reports of people in Kaohsiung falling ill after eating potato products in several large Facebook groups. Citing harm to businesses, public fear and a poor attitude after the offence, they asked the court for a heavy sentence.
Police arrested two minors on charges related to child pornography and privacy violations, accusing them of lifting the faces of underage girls from photos posted online, inserting them into other pictures and sharing the results on a social media platform. A parent was arrested on suspicion of neglecting supervision of a minor; all three went before a prosecutor.
A Singapore High Court assistant registrar ordered lawyer Deepak Natverlal to personally pay Choco Up $900 after his submissions misstated two real cases, one as holding the opposite of its actual ruling, and quoted a passage that appeared completely fictitious. He denied using AI; the court voiced doubts but said whether AI was involved was immaterial.
Who did it: HYBRID AI Hallucination Cases database
A Utah federal magistrate judge ordered plaintiffs' lawyers Tyler Ares and Amy Martz to pay defendants including Canyons School District $7,000 after their amended complaint cited non-existent cases. Counsel said AI used only to polish their writing had distorted real cases; the court was skeptical, noting they had already been admonished over AI use in another case.
Who did it: HYBRID AI Hallucination Cases database
Costa Rica's Family Tribunal ordered its ruling and an appeal authenticated by lawyer Carlos Salazar Solís sent to the Bar Association's prosecutor after the appeal quoted attorney-fee decree text that appears nowhere in the decree. The tribunal said the passage looked like a summary or copy of "some technological consultation" that "evidently had hallucination".