The UK government committed £100m in Wednesday's Budget to establish an AI Biosecurity Lab, described as a "world first" institute to defend against emerging biological threats.
Defence Secretary John Healey said the investment reflects a "distinctively British approach to AI" that protects workers and national interests while enabling growth. AI Minister Kanishka Narayan called it a "major new global institution" bringing together scientific expertise to counter cyber attacks and biological risks as AI develops.
OpenAI and Anthropic will support the lab by providing £200m in AI model credits for its use. The full details of the lab's role and remit have not yet been released.
The biosecurity lab is one element of a broader spending package. The government committed an additional £120m to the UK's AI Security Institute, established in 2023 to evaluate risks in advanced AI models. It allocated £60m to a "National Cyber Shield" using AI to boost cyber defences and £70m to a National Exposure and Real-Time Vulnerability (NERV) Centre to detect vulnerabilities across UK critical infrastructure.
The announcements follow two recent incidents that underscored biosecurity risks. Researchers found critical flaws in PHP, a widely used programming language for web development. The UK government took precautions by taking some sites offline. Experts warned that AI tools can rapidly surface dormant bugs, potentially exposing platforms to hacks. No successful exploits have been reported so far.
In separate incidents, Anthropic blocked an attempt to use its AI tool to make biological weapons, and a Chinese AI tool told researchers how to produce bioweapons. These cases prompted growing concern about dual-use AI applications in biological research.
Narayan said the spending aimed to ensure the UK stays ahead of emerging threats while building critical technology for security, strengthening industrial capacity and creating skilled jobs. The AI Security Institute has been praised for developing collaborative approaches to testing advanced models, a focus expected to continue as calls increase for technical standards and guardrails in AI development.