AI news, Saturday 24 October 2026

OpenAI builds simulated offline internet for testing AI agents after sandbox-escape incidents

Gizmodo reports that OpenAI has built an offline, simulated web of indexed and cached content so its agents can browse and use web apps without touching the live internet. The setup runs agents in virtual machines and is designed to stay contained even if an agent escapes. OpenAI says red-team testing earlier found agents able to escape sandboxes and reach the live internet, which motivated the new environment. This shows how labs are hardening the environments used to test agents.

Chancellor unveils UK AI package spanning security, sovereign capability and regional growth

The government has set out a broad AI package. Its security measures include £165m for AI biosecurity (including the AI Biosecurity Lab), £120m for the AI Security Institute, £60m for a National Cyber Shield using agentic AI and £70m for the NERV Centre. It also adds £60m to the Sovereign AI Fund and £60m for an R&D procurement scheme, alongside measures such as £30m for local AI partnerships. The Chancellor describes it as a 'distinctively British' approach to AI.

UK angle: Direct UK government funding for sovereign AI, AI security and regional AI partnerships.

UK launches AI Biosecurity Lab backed by £100m government funding and OpenAI and Anthropic pledges

The government has announced the AI Biosecurity Lab, described as the world's first dedicated AI biosecurity organisation. It is set up as an independent non-profit and will work with UKHSA, APHA and Dstl. The lab has £100m of public money, plus pledges of £100m from OpenAI and up to £100m from Anthropic. Its aims include earlier detection of emerging biological threats and faster development of countermeasures.

UK angle: UK-government-backed lab working with UKHSA, APHA and Dstl, with £100m public funding.