East Riding of Yorkshire Council is spending £60,000 on an AI tool for a 'no wrong door' approach to crisis support, plus £89,457 on devices for residents.

Jade Leung will leave her full-time roles as the PM's AI adviser and AISI chief technology officer, staying on as AISI vice-chair and AI Taskforce security adviser.

OpenAI, Anthropic and Hugging Face bosses told the UN that AI needs common risk-evaluation standards. Trump's tech adviser Michael Kratsios turned them down.

Australia's PM says an OpenAI agent got into a Services Australia Medicare statistics portal in June, and that OpenAI took until September to say so.

Insurers are weighing tighter cyber cover while they work out how to price agentic AI attacks. The Lloyd's Market Association is writing a model AI definition.

Unions at York, Lancaster, St Andrews, Reading and UEA question Turnitin's licence terms. The firm says no update will take effect before 1 September 2027.

Anthropic says around 950 Claude agents combed DNA data for 21 hours and flagged a new reverse transcriptase system with CRISPR-like repeats. Its role is unknown.

The CMA plans to let AI assistants appear on Google's UK search choice screens, but eligibility criteria are still under consultation until 9 October.

The Alan Turing Institute says doubts over existential AI threats are no excuse for delay, and sets out five risks plus a £2m security research programme.

A CSconnected report counts 3,140 compound semiconductor jobs in Wales, paying about £66,000 on average. Welsh firms say AI is creating roles, not cutting them.

WPP's new East London studio combines AI production, shoot and edit space for fast social ads, the first of a global roll-out under Cindy Rose's turnaround plan.

Anthropic says Opus 5.5 matches Fable 5.1 on most work for 40% less than Opus 5; OpenAI halves API prices for GPT-6 Sol and Luna. What it means for buyers.