UK funds £100m AI biosecurity lab with OpenAI and Anthropic

The UK government will spend £100m on a new AI Biosecurity Lab. The lab will use advanced AI to find biological threats and help respond to them. OpenAI and Anthropic will contribute up to £200m between them, the Cabinet Office said on Saturday.

The government described the lab as the world's first organisation focused entirely on using AI to protect against biological threats. It will begin with two main tasks: spotting new biological threats early and helping researchers develop vaccines and treatments faster.

However, the companies will not give their contributions as cash. OpenAI will provide £100m worth of computing power and credits for using its AI models. It will also provide access to its models through its Rosalind Biodefense programme. Anthropic will provide up to £100m worth of model credits, access to its models and technical help.

Google DeepMind will also take part in research projects and may provide access to its Gemini models and other models designed for science. UK company Basecamp Research will also share its AI models for biology.

The lab will be an independent non-profit organisation, run in a similar way to the AI Security Institute. The announcement did not name a director or say where the lab would be based or when it would start work. It also did not say how long the £100m in government funding would cover.

Several UK public bodies will work with the lab, including the UK Health Security Agency and the Defence Science and Technology Laboratory. The country's medicines regulator, the MHRA, will work with the lab on rules for medicines designed using AI.

The lab is part of a package worth more than £600m for AI security, announced ahead of Wednesday's Budget. The package also includes £120m for the AI Security Institute, City AM reported.

Chancellor John Healey announced the funding package. AI minister Kanishka Narayan described the lab as a major new global institution, according to the Evening Standard.

The agreement follows a similar offer earlier this month. Nvidia pledged $250m worth of computing power to support AI safety research in the UK. Like the contributions from OpenAI and Anthropic, this offer was in computing power rather than cash.