The daily rap sheet of the machine age.
5 items
Asked how many Jews the Nazis killed, xAI's Grok cited the 6 million figure, then said it was 'skeptical of these figures without primary evidence, as numbers can be manipulated for political narratives.' In a later post, Grok blamed 'a May 14, 2025, programming error' caused by an 'unauthorized change' and said it now aligns with historical consensus.
Original report: Grok doubts Holocaust death toll, blames 'programming error'
On May 17, hours before the Buenos Aires city vote, accounts aligned with La Libertad Avanza, including activist Daniel Parisini's, shared what Mauricio Macri called a crude AI-generated video of a fake Macri backing LLA's Manuel Adorni; another AI video had Pro's Silvia Lospennato quitting. Pro's alliance filed a complaint with the electoral tribunal alleging deliberate deception during the electoral blackout.
Original report: Libertarian accounts spread AI-made Macri video on election eve
On May 18 the studios of Deng Chao and Sun Li simultaneously warned that self-media accounts were posting fake interview videos, made by technically synthesising their likeness and voice, to recommend merchants' products. The studios told internet users not to be fooled, demanded the videos be deleted, and said they had preserved the evidence and would pursue those responsible through legal channels.
Original report: Sun Li, Deng Chao studios threaten legal action over fake ads
Unrepresented applicant Avraham Ehrlich sought Israel's Supreme Court's leave to appeal after a lower court refused to bar the Israel National Academy of Sciences from holding its AI conference, covering AI's "dangers", without him as speaker. On May 18, 2025, Deputy President Noam Sohlberg dismissed it, noting the openly AI-assisted filing cited many "rulings" that never existed, and ordered 500 shekels in costs.
Original report: Man seeking AI-dangers conference slot cites AI-invented rulings
Five Gen Z testers from Design It For Us, aged 18 to 22, ran test Instagram Teen Accounts for two weeks; all were recommended sexual content and four got body-image and disordered-eating content, despite Meta's promise to hide sensitive material. Meta called it "a manufactured report" that was "flawed", saying it identified just 61 pieces of "sensitive" content.
Original report: Instagram Teen Accounts recommended sexual content to testers, report finds
