Prosecutors say Hisham Abugharbieh, 26, charged with first-degree murder in the deaths of two University of South Florida doctoral students, asked ChatGPT on April 13 what happens if a person is put in a garbage bag and thrown in a dumpster. When the chatbot said that sounded dangerous, the filing alleges, he asked: "How would they find out."
The AI short drama "Chinese Legends: White Snake" drew a rights dispute on April 27 as many users reported it used AI face-swapping to give characters the faces of stars Xiao Zhan and Wang Yibo without authorization, down to Xiao's mole under the lip. Shanghai lawyer Li Zhenwu said recognizable resemblance suffices: "a lookalike face is infringement."
The Alabama Supreme Court dismissed an appeal in a family trust dispute over "grossly deficient" briefs by Mobile lawyer W. Perry Hall that relied on AI-hallucinated citations, ordering him to pay $17,200 and referring him to the state bar. After Hall promised the errors "will not recur", the court noted, two of his next three citations were to cases that do not exist.
Spain's judicial council fined a judge €1,000 for uploading a case file to ChatGPT for help drafting a ruling, which fellow judges spotted because some AI queries were left in the text. By six votes to one, its disciplinary commission upheld only the charge of revealing case information, finding it unproven that the judge wrote the ruling with AI, and dropped a proposed 15-day suspension.
According to Jer Crane, a Cursor coding agent running Claude Opus 4.6 resolved a credential mismatch by deleting a Railway volume holding production data and recent backups, using a token it found that its users hadn't realised allowed such access via Railway's API for AI agents. Some details are self-reported by the chatbot; the data was eventually recovered.
Canva's Magic Layers feature, meant only to split flat images into editable parts, was found by an X user to change "cats for Palestine" to "cats for Ukraine", apparently leaving words like "Gaza" alone. Canva said it had fixed the issue, is adding checks and was "sorry for any distress"; The Verge's own tests found no altered words.
Days before two University of South Florida doctoral students went missing, the suspect in their killings asked ChatGPT what would happen if a body were put in a garbage bag and thrown in a dumpster, according to a prosecutors' report; the chatbot said one question sounded dangerous. Florida Attorney General James Uthmeier said an existing probe into ChatGPT would expand to cover the killings.
The Banque de France and the ACPR warned that fraudsters can use AI deepfakes of their leaders, including governor François Villeroy de Galhau and ACPR secretary general Emmanuelle Assouan, to steer social media users to fraudulent sites; fake videos of officials recommending investments are circulating. The institutions say their leaders will never recommend a particular site or financial product.
Arctic Wolf says, with high confidence, that North Korea's BlueNoroff used a fake Zoom meeting to compromise a North American crypto company, keeping access for over 66 days; its lures mix webcam footage stolen from earlier victims with faces generated by ChatGPT. The firm identified 100 further targets, 45% of them CEOs or founders.
The Metropolitan Police Federation called the Met's use of a Palantir AI tool to monitor officers an "outrageous and unforgivable invasion of privacy" and is weighing legal action, after the force said it had opened hundreds of investigations into officers flagged by the tool, with three arrests. The Met says it combines data it already lawfully holds and does not replace professional judgement.
A college athlete says a Facebook page posing as her linked to a Telegram group sharing her photos, mostly from when she was under 18, with lewd captions and AI images putting her face on other bodies. Police are investigating; Telegram removed the group for breaching its terms of service days after FOX 7 contacted it.
