U.S. District Judge William Alsup ruled that Anthropic made 'fair use' of authors' books to train its Claude model, calling the training 'exceedingly transformative'. But he found that copying and storing more than 7 million pirated books in a 'central library' infringed the authors' copyrights, and ordered a trial in December to determine how much Anthropic owes.
France's consumer protection authority, the DGCCRF, found Tesla engaged in misleading business practices, including over the 'fully autonomous driving capacity' of its vehicles, plus incomplete sales contracts and late refunds. Tesla has four months to comply or face fines of 50,000 euros a day. Its Full Self-Driving option, sold in France for 7,500 euros, is not yet active there.
Business Insider found Scale AI routinely used public Google Docs to track work for clients including Google, Meta and xAI, leaving documents marked 'confidential' open to anyone with the link, along with spreadsheets of thousands of contractors' names and private email addresses, some flagged for suspected 'cheating'. Scale said it is investigating and has disabled public sharing.
National deputy Rodrigo de Loredo of the Unión Cívica Radical posted an AI-generated video in which Córdoba Governor Martín Llaryora says things he never said, using it to criticise tax increases. A social media commentator called it a 'dangerous use of AI'; De Loredo responded that the video made clear it was AI-made.
Cannes Lions is investigating DM9's Grand Prix-winning campaign for Consul after an anonymous complaint alleged distorted edits of a TED Talk by North Carolina state senator DeAndrea Salvador. DM9 acknowledged 'a series of failures' in its videocase and announced an AI ethics committee; Salvador has sued Whirlpool and Omnicom, accusing them of manipulating her image, voice and speeches with AI.
Who did it: HYBRID AI Hallucination Cases database
A Canadian court ordered an applicant's counsel in an immigration case to personally pay $100 in costs after he relied on cases that did not exist. It took four court directions before he admitted generative AI was used; he said he relied in good faith on Visto.ai. The judge said the real issue was his failure to disclose the AI use.
Who did it: HYBRID AI Hallucination Cases database
Paraná's Court of Justice unanimously upheld a bad-faith litigation fine of 5% of the case value against two plaintiffs whose lawyer cited precedents that, the court found, did not exist and were invented by AI, along with a referral of the lawyer to the Paraná bar. The lawyer said an AI research tool generated them in error.
An Ontario senior lost more than $15,000 after clicking a Facebook ad featuring a deepfake of Premier Doug Ford promoting a mutual fund account, Guelph police said. The ad led him to a chat page where he was persuaded to 'invest' over the following week through e-transfers and credit-card payments.
Doctors are using AI note-taking software that does not meet minimum standards to record and transcribe patient consultations, according to a Sky News investigation. A 9 June letter from NHS England's national chief clinical information officer warned that non-compliant ambient voice technology "still being widely used" risks clinical safety and data protection breaches.
