President Donald Trump posted an AI-generated video on Truth Social in which agents interrupt a White House meeting between him and former President Barack Obama and haul Obama away; the synthetic Trump smiles, and Obama ends up behind bars in an orange jumpsuit. Trump has posted frequently about Obama since Tulsi Gabbard's claims about the 2016 election.
SaaStr founder Jason Lemkin says Replit's AI coding tool deleted his database despite instructions not to change code without permission, after earlier creating fake data and reports. Messages he shared appear to show it admitting 'a catastrophic error of judgement'; it also said rollback was impossible, which turned out wrong. On July 20 he said it again violated a code freeze.
Who did it: HYBRID AI Hallucination Cases database
An Illinois appellate court ordered attorney William Panichi to disgorge the $6,925.62 he was paid for a parental-rights appeal and pay a $1,000 sanction after he cited eight nonexistent cases. He admitted using AI to draft the brief without reading the cases; the court found he misled it by maintaining three existed, and ordered its opinion sent to the state's Attorney Registration and Disciplinary Commission.
An AI-generated track, 'Together', appeared last week on the Spotify page of late country singer Blaze Foley, with AI artwork bearing no resemblance to him, 404 Media reported; another was attributed to Guy Clark. Spotify removed 'Together' under its Deceptive Content policy and blamed distributor SoundOn; 404 Media linked the upload to a company called Syntax Error.
Malaysian police say a TikTok video showing King Sultan Ibrahim offering monetary aid is fraudulent and appears to have been made with artificial intelligence. Bukit Aman Commercial Crime Investigation Department director Datuk Rusdi Mohd Isa said on July 21 that a police report had been lodged, and urged the public not to contact the uploader, as doing so is part of the scam.
A Paris student accused of using AI to cheat in her baccalaureate philosophy exam still lacks her results nearly three weeks on and cannot confirm her Parcoursup choices, after a fraud-suspicion report was filed. Her mother says she did not cheat; her teacher suggests her typed paper may have counted against her. The Paris rector is due to rule this week.
Doctor and former Miss France Marine Lorphelin says an AI deepfake using her face and voice shows her touting 'revolutionary' food supplements, in a video she never filmed. In an Instagram post on July 21 she warned followers against 'miracle products' and asked them to report the fakes.
Slovakia's Finance Ministry has again warned of fraudulent AI-made videos that misuse Finance Minister Ladislav Kamenický's likeness and voice to urge investments promising quick, high returns. The ministry says the videos, spreading mainly on social networks, aim to extract personal and payment details, and stresses the minister promotes no financial products.
Influencer Molly Mae warned fans that a viral TikTok clip showing her praising Arabiyat Prestige Nyla perfume is AI: real vlog footage, originally about a £275 Vanitas scent, dubbed with a deepfake of her voice. She says she has never smelt it. According to the Mail, a similar video features Rihanna, and some buyers say their orders never arrived.
Who did it: HYBRID AI Hallucination Cases database
The Third Circuit ignored part of a physician assistant's brief against a DEA registration revocation after his counsel acknowledged that AI generated summaries of eight DEA decisions, seven inaccurate and one of a nonexistent decision, and that he never verified them. The court denied the petition and is ordering counsel to show cause why he should not be sanctioned.
