The Independent Publishers Alliance, joined by the Movement for an Open Web and Foxglove, has complained to the European Commission that Google abuses its market power in online search by using publisher material in AI Overviews, costing publishers traffic, readership and revenue. The groups seek an interim measure; Google says it sends billions of clicks to websites each day.
Tesla posted a video of its Full Self-Driving (Supervised) system navigating Madrid traffic, captioned 'Pending regulatory approval'. Madrid city mobility delegate Borja Carabante said the vehicle lacked the required permit, which the traffic authority issues with a favourable city report, and the DGT has asked Tesla for explanations. Tesla says the system handled the scenario safely.
Numerous outlets reported that DeepSeek had apologised for its AI improperly linking Wang Yibo to the corruption case of Li Aiqing. Huxiu found no such apology on any official DeepSeek channel; when it asked AI models about the affair, they all confirmed the apology, citing the false reports.
MCA complaints chief Michael Chong says scammers are using AI deepfakes of Prime Minister Anwar Ibrahim, Robert Kuok, Tengku Maimun Tuan Mat and Tony Fernandes to push fake investment schemes; he puts losses to these scams at RM2.11 billion last year. An MCA sting saw RM1,100 'invested', followed by a demand for a US$250 'processing fee'.
Rapper Potap, now performing as Slavic Balagan, posted an AI-generated teaser for a clip showing singers Tina Karol and Svitlana Loboda as opossums in a 'fight'. Karol replied that an artist out of ideas 'starts to parasitise on other people's faces', called the provocation 'cheap vagrancy' and signed off 'Adios Amigo'.
TikTok videos showing hundreds of people chanting 'Hugo Motta, traidor' at protests against the Chamber of Deputies president and Congress were made with Google's Veo, fact-checker Comprova found; one had 552,000 views. The Veo watermark was visible, but posts on an account styled 'Notícias do Brasil e do mundo' carried no clear AI disclosure. No such street protests are on record.
Who did it: HYBRID AI Hallucination Cases database
In an intimate images damages dispute, BC's Civil Resolution Tribunal found both self-represented parties relied on an AI-hallucinated version of section 92 of the Civil Resolution Tribunal Act, one quoting the entire fabricated section on monetary penalties. Vice Chair Eric Regehr said there was 'no way a human being' could make that error, and declined to answer AI-concocted arguments.
In an Intellectual Property Office appeal over the sign ProHealth, litigant in person Dr Mustapha Soufian admitted drafting the case with help from ChatGPT; appointed person Phillip Johnson pointed out numerous citation errors. Respondent's trade mark attorney Victor Caddy was also criticised for genuine citations that appeared not to support the submissions, telling the hearing: 'I cannot actually remember that now.'
State Bank of India issued a caution notice on X over a deepfake video falsely claiming an AI-based investment platform had launched with SBI's support, in association with the Government of India and a few multinationals, promising high returns. SBI says it endorses no schemes promising unrealistically high returns and urges checks via its website, social media handles or branches.
