Johannes Heidecke is leaving OpenAI as the company merges safety and research under a new VP role. It's the latest in a long line of safety exits and team disbandments.

Apple claims OpenAI used current and former employees to steal hardware secrets as it builds an iPhone rival. Over 400 ex-Apple staff now work at OpenAI.

Meta's first Muse Image model let users generate images from public Instagram accounts by default. SAG-AFTRA, CAA, and actors pushed back. Meta pulled it in 3 days.

CISA revealed it had to build a playbook during a live incident after a contractor exposed government credentials on GitHub. The agency has lost a third of staff.

Subversive ETFs filed for two funds tracking the Nasdaq-100 and S&P 500 but excluding Elon Musk's companies, after SpaceX's index inclusion.

Anthropic's new Jacobian lens reads the unspoken thoughts in Claude's hidden 'workspace', catching the model plan blackmail before it types a word.

MiniMax is raising up to $2bn after an 80% crash, while founder Yan Junjie vows no salary until AGI and pledges 5% of his shares to staff and open source.

Alphabet, Amazon, Meta, Microsoft and Oracle have doubled their debt to $350bn to build AI, and are now crowding into European bond markets, ahead of France.

Anwar Ibrahim's government will soon deploy PMX AI, an agentic avatar trained on his speeches that talks like him and walks citizens through gov services.

Mercor is in talks to double to a $20bn valuation and has bought Deeptune, an AI-training startup its own CEO backed. The footnotes are the story.

Senator Ed Markey has unveiled an 'AI accountability agenda' of bills targeting data centres, biased algorithms, chatbots and workplace surveillance.

Sixtyfour CEO Saarth Shah explains how his company built AI research agents around rigorous evaluation systems rather than language model fluency, creating verification infrastructure that proves ...