Nvidia Allocates $250 Million for AI Safety Research in Britain

Nvidia has announced a compute donation valued at $250 million over five years, approximately $50 million annually, to support AI safety and security research in Britain through UK-based cloud partners, according to the Nvidia newsroom.

Beneficiaries include the "BOLD" lab led by the University of Oxford, the "SOFAIR" lab led by University College London, and Imperial College London, according to the same source.

This research funds areas including safe agents, open model security, and threat mitigation, tracks linked to the growing use of AI agents in sensitive tasks.

The announcement includes statements from UK AI Minister Kanishka Narayan and Business Secretary Jonathan Reynolds, at a time when Britain is seeking to strengthen its safety research infrastructure.

What do these terms mean?

Compute donation: Providing processing power or runtime on data centers for free instead of granting cash, which is the most critical currency in AI research.

Safe agents: AI agents designed to operate with controls that prevent unintended or harmful actions.

Open model security: Protecting publicly available models from misuse or harmful modification.

Threat mitigation: Measures that reduce the likelihood of damage or limit its impact when it occurs.