Nvidia announced this Wednesday (14) a commitment of US$ 250 million in computing capacity to support artificial intelligence safety research in the UK over the next five years. The initiative includes projects from the universities of Oxford and University College London (UCL), in addition to Imperial College London.
The amount does not represent a transfer of money to the institutions. The chipmaker will provide access to processing resources valued at US$ 50 million per year, intended for the development and evaluation of AI systems, especially agents capable of executing tasks autonomously.
Most of this capacity will be made available to the British AI safety research community. A portion will serve specific projects, but Nvidia did not disclose individual amounts for the participating laboratories.
Resources will be provided by cloud providers in the UK
The program will use Nvidia's cloud computing partner network, with infrastructure installed on British territory. The choice of suppliers, the technical architecture and the commercial conditions have yet to be defined, according to the company's statement.
Among the identified beneficiaries is the British Open-ended Learning and Discovery Lab (BOLD), led by the University of Oxford, which investigates methods to develop systems capable of learning, adapting their behavior and interacting with people.
The Science of Fundamental AI Research Lab (SOFAIR), coordinated by UCL, works on the development of open artificial intelligence architectures. Imperial College London, in turn, will concentrate part of the research on safety engineering and robotics.
BOLD and SOFAIR already existed before the commitment announced by Nvidia. In June, the British government had announced funding of up to £60 million for the two laboratories over six years. The new program expands the available computing support, without replacing that public funding.
The research will also cover protection mechanisms for autonomous agents, safety of open models and threat testing in controlled environments. The initiative foresees the development of reproducible evaluation parameters, including work with Nvidia's OpenShell and Sentry technologies.
The commitment occurs as the UK seeks to expand its capacity to assess the risks of advanced AI systems. The country already maintains the AI Security Institute, a public institution dedicated to investigating risks and evaluating artificial intelligence models.
