OpenAI and Anthropic Back Britain’s AI Biosecurity Lab

LONDON, Oct. 24 — BritainBritain will put £100 million into a new AI Biosecurity Lab, an independent nonprofit meant to use frontier AI to detect biological threats earlier and to speed the development of vaccines and treatments, and OpenAI and Anthropic have pledged up to £200 million more in model credits. The Cabinet Office announced the lab a minute after midnight on Saturday, London time, and HM Treasury set out the wider package of AI spending around it at 6 a.m.

By the Cabinet Office’s account, OpenAI will provide £100 million in compute and model credits and trusted access to its most capable relevant models through its Rosalind Biodefense program.Rosalind BiodefenseAn OpenAI program launched in May that, in the company’s words, expands “trusted access to GPT-Rosalind for vetted developers and U.S. government partners” working on biodefense, public health and pandemic preparedness. GPT-Rosalind, introduced in April, is OpenAI’s reasoning model for life-sciences research. Anthropic will provide up to £100 million in model credits, trusted model access and dedicated technical support. Google DeepMind and Basecamp Research, a London company that builds AI models for biology, join as research partners, with no sum attached.

“Our new AI Biosecurity Lab will be a major new global institution, bringing together world-leading scientific expertise to protect people from emerging biological threats and ensure Britain leads the world in AI safety and security,” Kanishka Narayan, the AI minister, said in the Treasury’s release. The lab’s own page is plain about the threat it answers: “increasingly capable AI systems also create new biological risks and could lower barriers to harmful activity.”

For the two American companies, the commitments the government announced are in credits, model access and technical support, and they place their models inside a biodefense institution the government is establishing. It joins a national initiative with the UK Health Security Agency, the Animal and Plant Health Agency and the Defence Science and Technology Laboratory, and is to work with the medicines regulator to “inform the regulation of AI-designed medicines.” The pledges come weeks before the stock-market listing Anthropic plans for mid-November, Reuters reported, citing people familiar with the matter, who cautioned that the plan and the dates could change, and as OpenAI nears a new private funding round, according to sources CNBC cited. Neither company had posted about the lab on its own news page by early Sunday; the terms are the government’s.

The lab is one line in a longer list. Its £100 million is part of £165 million for a National AI Biosecurity Initiative, whose other £65 million goes to bio-surveillance and to medical countermeasures for priority pathogens. The AI Security Institute, which tests new AI systems, gets £120 million more.

Two cyber programs get £130 million between them: £70 million for a National Exposure and Real-Time Vulnerability Centre, or NERV, to deploy AI tools to “identify cyber vulnerabilities across our critical infrastructure and the economy,” and £60 million for a National Cyber Shield that will use agentic AI. The Sovereign AI Fund, which invests directly in British AI companies, gets £60 million more, and a UK Advanced Chip Design Academy gets £18 million.

Some of the money is meant for companies. The Cyber Shield includes a £25 million challenge fund for British AI and cyber-security innovators, and another £60 million goes to Sovereign AI’s R&D Procurement Scheme, to deliver solutions to challenges “across the health and care system and in AI security, assurance and safe adoption.”

Mr. Narayan cast the package as a matter of independence. “We cannot afford to rely entirely on technology built overseas,” he said. The lab at its center will launch with up to £200 million committed by two American companies.

The Treasury’s release on the wider package does carry a total, “£600m+,” but not in the government’s voice: it comes from Julian David, chief executive of techUK, the technology trade association, who called it “a strong £600m+ package” in a statement the Treasury included.The SumThe sums of public money the release names add up to £621 million, by this paper’s arithmetic, plus £10 million a year for work on how AI will reshape jobs. It gives a period only for that line and for the £23 million, over three years, for the AI Economics Institute. A new AI cyber-security center in Manchester has no figure. The Cabinet Office’s own summary of the lab, “up to £300mn of government funding and company commitments,” adds the companies’ credits to the government’s £100 million.

The Treasury says the chancellor of the Exchequer, John Healey, will announce the package in his Budget, which he presents to Parliament on Wednesday, Oct. 28. Mr. Healey said he was “setting out a distinctively British approach to AI,” one that “doesn’t leave workers behind and defends our national interest.”

Much is unsettled. The lab’s page names no director, no site and no opening date, and neither page says over how many years the companies’ credits run.

Daily Cortex is written with AI models, including Anthropic’s Claude, and fact-checked with OpenAI’s GPT models.