Democratic Senators Elizabeth Warren and Ron Wyden sent letters to Google, Anthropic, Microsoft and OpenAI asking whether their AI partnerships act as "de facto mergers" that bypass antitrust scrutiny, citing Google's 14% stake in Anthropic after a $3 billion investment. The companies were given until April 21 to answer questions on computing access, staff overlap and possible acquisitions.
Former Facebook global public policy director Sarah Wynn-Williams was due to tell a Senate subcommittee that Meta began briefing the Chinese Communist Party on AI as early as 2015 to help China outcompete American companies. Her prepared remarks draw a "straight line" to Chinese military AI built on Meta's Llama. Meta called her testimony "riddled with false claims."
Police in Johor arrested a 16-year-old boy suspected of using AI to turn schoolmates' and alumni's social media photos into pornographic images, which were sold at RM2 each. Police received eight reports. Deputy Digital Minister Teo Nie Ching said 30 to 40 people were believed to be victims, mostly alumni but including students as young as 14.
Microsoft faces criticism over its cloud and AI support for Israel's military, highlighted by employee protests during its 50th anniversary celebrations. According to an Associated Press report, the military's use of Microsoft and OpenAI AI tools in March 2024 was nearly 200 times its pre-October 7, 2023 level, and it uses Azure to process mass-surveillance intercepts.
SentinelLABS says AkiraBot, a spam framework, used OpenAI's gpt-4o-mini to write messages tailored to each site, promoting a dubious SEO service, and used CAPTCHA-bypass tools to get them through. It targeted more than 400,000 websites and spammed at least 80,000 since September 2024. OpenAI said the API key involved has been disabled.
Former Sinn Féin president Gerry Adams says his books were used without his permission and has passed the matter to his solicitor. An Atlantic investigation found Meta may have trained Llama on millions of pirated books and papers from LibGen. Meta says its use of information to train AI models is consistent with existing law.
Nelson Lin, a lawyer representing himself in a federal suit against a New York Family Court judge and others, cited Hernandez v. Spitzer to argue that federal courts routinely hear such claims. Judge John G. Koeltl wrote that "that case does not exist." On March 6, 2026 the court dismissed the action without prejudice for lack of subject-matter jurisdiction.
A woman in Salento has filed a complaint after an Instagram deepfake of Mediaset chief executive Piersilvio Berlusconi, promoting an investment platform, led her to pay nearly €12,000 to strangers. Asked for further money, including a €7,300 "tax" to release supposed €28,300 gains, she stopped at the bank. Prosecutors will decide whether to open a case.
