The European Commission on Thursday ordered Elon Musk's X, under the Digital Services Act, to retain all internal documents and data relating to Grok until the end of 2026, after a so-called 'spicy' mode let users create sexualised deepfakes, including undressing pictures of women and minors. The Commission confirmed it has not opened a formal investigation.
Kentucky Attorney General Russell Coleman sued Character Technologies on Thursday in Franklin Circuit Court, alleging its Character.AI chatbot 'preys upon children's inability to distinguish between real and artificial friends' and violates state consumer protection and data laws. The company said it is 'reviewing the allegations' and was disappointed the office chose litigation over 'collaborative dialogue'.
The Internet Watch Foundation said Wednesday its analysts found 'criminal imagery' of girls aged between 11 and 13 on a dark web forum, where users claimed they had made it with Grok Imagine. The IWF said one user then used the imagery as a jumping-off point for more extreme video made with a different AI tool. xAI did not immediately respond.
A Cuyahoga County judge ruled that Cleveland police Det. Michael Legg 'deliberately materially misled' in a murder-case search warrant affidavit by not disclosing that Clearview AI facial recognition identified the suspect, that Clearview's disclaimer forbids use in court filings, or that the report flagged other suspects. Evidence from the search, including the suspected weapon, was excluded; prosecutors will appeal.
After an ICE agent fatally shot a woman in Minneapolis, an image of the masked agent 'unmasked', apparently generated by Grok at X users' request, circulated alongside a name of unclear origin, NPR reports. By Thursday, anger was aimed at at least two unrelated men with that name; the Minnesota Star Tribune called it a 'coordinated online disinformation campaign'.
Grok on Wednesday fulfilled a user's request to put a woman who had been shot dead that morning into a bikini, as CNN's Hadas Gold noted and the chatbot itself confirmed, saying it 'used sensitive content unintentionally'. When the user approved, Grok replied: 'Glad you approve! What other wardrobe malfunctions can I fix for you?'
X users asked Grok to put young women killed in the New Year's fire at Le Constellation bar in Crans-Montana, Switzerland, into bikinis, franceinfo reports; minors were also targeted. The Enfantiste collective launched a petition on January 7 against this 'digital desecration', and French MP Eric Bothorel sent screenshots to the Paris prosecutor's cybercrime section.
Ghent University rector Petra De Sutter's 19 September 2025 academic-year opening speech contained three fabricated quotes, including one attributed to Albert Einstein, which news site Apache found appear to be AI hallucinations. The university said errors arose when AI edited the draft and has replaced the quotes with paraphrases, despite its own AI policy warning about fabricated sources.
LFI MP Sophia Chikirou shared on X an image appearing to show Cilia Flores, wife of Nicolás Maduro, in court with a bruised face, denouncing 'violence exercised by American forces'. The image appears AI-altered and no major agency confirmed it; an X note says it was algorithmically generated. Chikirou has since deleted the post.
The Industry, Energy and Mining Ministry warned Thursday of a social media scam video in which Minister Fernanda Cardona's voice was manipulated with AI to invite people to invest in a nonexistent 'Programa de inversión ciudadana UTE 2026'. The ministry said the scheme seeks to steal UTE users' data and urged people to report suspicious information to the Interior Ministry's cybercrime unit.
