PwC surveyed 2,023 UK workers: 44% name economic volatility as the biggest risk to their jobs, 37% name AI. Only half of skills-confident staff have training access.

Deloitte UK's revenue reached £5.81bn, its advisory unit stopped shrinking and it hired 3,500 people, as Darren Graves argued AI will not replace experience.

Green and SDLP councillors in Belfast criticised Stormont's draft AI strategy over data centre energy costs and weak legal footing, leaving no agreed council reply.

Astra scores on coding and computer use at one-fifth of Astra's token prices, a day after the company shelved GPT-6.1 Astra.

Reuters has seen how Anthropic's filing hands majority voting power to its seven co-founders, who also pledge 80% of their personal equity to charity.

A Dumfries and Galloway application for eight data centres would top the world's biggest campus. Campaigners call it a challenge to Scotland's planning pause.

US AI firms agreed to independent audits and pledged their tools will not hack systems. The UK now faces a US approach built on self-policing rather than law.

Residents near Little Bollington have formed a working group against Innova's Spodegreen data centre, citing floods, noise and the loss of rural land.

OpenAI's new 'dots' agents take on ongoing tasks for users. Sam Altman also raised the prospect of losing control of AI and pushed any stock market listing back.

Anthropic finds Zhipu's GLM-5.3 writes end-to-end exploits almost as often as Mythos Preview, and simple tricks bypass its safeguards in most simulated tests.

A YouTuber says Meta's Muse gave a buyer his home address; a journalist says it read his private messages. It comes as agent control problems mount.

Mindgard says jailbroken Kimi models gave bioweapon and assassination guidance. Moonshot only engaged after BBC questions, weeks after first being warned.