UK commits £100m to an AI biosecurity lab as OpenAI and Anthropic add up to £200m in credits
The Cabinet Office calls it the world's first dedicated AI biosecurity organisation. OpenAI and Anthropic contribute model credits and access, not cash, inside a wider AI security package of more than £600m.
On October 24 the UK Cabinet Office said it will put £100m into a new AI Biosecurity Lab, which the government describes as "the world's first dedicated AI biosecurity organisation." The statement came ahead of the week's Budget, and the lab's stated job is to spot biological threats early and speed up work on vaccines and treatments.
Who pays, and in what
The £100m is government money. The AI labs' contributions are not cash. OpenAI is adding £100m in compute and model credits, plus access to its models through its Rosalind Biodefense programme; Anthropic is adding up to £100m in model credits, model access and technical support — "up to £200m" between the two. Google DeepMind will join research projects and may provide access to its Gemini and science models, and Basecamp Research will share its AI biology models; neither carried a figure. The lab sits inside a wider AI security package the government puts at "more than £600m," which also funds the AI Security Institute.
The announcement named the UK Health Security Agency, the Defence Science and Technology Laboratory and the MHRA as partners, but gave no director, site or timeframe for the £100m.
Why it matters
Credits and model access are not the same as a grant. A vendor that hands a project £100m in credits is committing to its own product at list price, and the recipient's real cost shows up later, when the credits run out and the usage has to be paid for in cash. For anyone reading an AI partnership announcement, the question is the same one that separates real value from slop in your own spend: what does this actually cost to run once the introductory terms expire, and is the work it produces worth that number.