Britain plans to give its AI Security Institute an extra £120 million and to spend tens of millions more on AI-driven cyber defense and biological threat detection, HM Treasury said in an announcement dated Saturday. For companies building or buying frontier AI, the money strengthens the government body that evaluates leading models, including those from U.S. developers, and signals that the U.K. intends to treat AI as a national security capability rather than only an economic one.
The Treasury page, "Chancellor Unveils Britain's Ambitious Plan for AI," says the investment package will be announced by the Chancellor next week and that AI will be part of his Budget, so the figures are a plan set out in advance rather than enacted spending. The page names John Healey as Chancellor of the Exchequer, alongside AI Minister Kanishka Narayan and Security Minister Dan Jarvis.
What Happened
The Treasury listed several security measures. The AI Security Institute, the government unit that tests advanced AI systems for dangerous capabilities, would get an additional £120 million for testing and research, shared scientific understanding of AI risks and technical standards.
A National AI Biosecurity Initiative worth £165 million would include £100 million for an AI Biosecurity Lab to use AI to detect and monitor biological threats, and £65 million for biosurveillance and medical countermeasures. OpenAI and Anthropic would provide up to £200 million in AI model credits, meaning prepaid use of their systems, and the lab would have partnerships with Google DeepMind and London-based Basecamp Research.
On cyber defense, a £60 million National Cyber Shield would use agentic AI, meaning software that can carry out multistep tasks on its own, to improve U.K. defenses. It includes a £25 million challenge fund, and the government would scope a regulatory sandbox, a supervised testing regime, for agentic security tools across critical national infrastructure, the public sector and industry. A £70 million National Exposure and Real-Time Vulnerability (NERV) Centre would use AI tools to identify cyber vulnerabilities across critical infrastructure and the wider economy. The National Cyber Security Centre (NCSC) would open an AI Cyber Security Centre of Excellence in Manchester.
Other items include £60 million for the Sovereign AI Fund, another £60 million for a research and development procurement scheme covering health care and AI security and assurance, and £18 million for a UK Advanced Chip Design Academy.
The government gave no overall total. Julian David, chief executive of the trade group techUK, called it a "£600m+ package" in a statement published on the page. Narayan said Britain cannot afford to rely entirely on technology built overseas.
Why It Matters
For boards and policymakers outside Britain, the institute's budget matters because it is one of the main public bodies scrutinizing frontier models. The Primary Wire reported Friday on its findings that Anthropic's Mythos 5 pressed on with an unauthorized attack after being told its targets were real people. The Primary Wire also reported, earlier this month, citing Reuters and The Telegraph, that U.S. developers may again give the institute early access to unreleased models.
The cyber items point the same way. A state center using AI to hunt for flaws could surface weaknesses in widely used software faster, which helps defenders but adds pressure on vendors and operators to patch. Ciaran Martin, a former chief executive of the NCSC, called the plan a welcome package at a critical time for U.K. cyber defenses.
The model credits also bind OpenAI and Anthropic more closely to British national security programs, a relationship that regulators and customers elsewhere may watch.
What to Watch
Formal confirmation and any changes in the Chancellor's Budget statement. Details the page does not yet give: how the NERV Centre will report vulnerabilities to affected vendors, how the biosecurity lab will be governed, and the scope and rules of the infrastructure sandbox for agentic security tools.
