About the GovAI Risk Management Team

The GovAI Risk Management Team currently consists of Jonas Freund (team lead), Noemi Dreksler, Sophie Williams, Aidan Homewood, Zaheed Kara, and Tom Reed.

Our goal is to strengthen risk management practices at frontier AI companies, with a special focus on safety frameworks.

We conduct research, and advise companies (e.g. Anthropic, OpenAI, and Google DeepMind) and governments (e.g. the EU AI Office and UK AISI).

At the moment, our main priority is helping companies update their safety frameworks in response to the EU GPAI Code of Practice.

Although we also publish papers, the majority of our work consists of private advising engagements.

Decision-makers at frontier AI companies and in government often need to make high-stakes decisions under time pressure and uncertainty.

Academic researchers could think more about the underlying issues, but the findings are often not interesting or novel enough to be publishable. Academics might also lack the necessary context to do work that is immediately relevant to decision-makers.

We try to bridge that gap. We conduct independent and rigorous research that informs important decisions at companies and in government.

Here are some of our recent publications:

• Assessing risk relative to competitors: An analysis of current AI company policies (by Sophie Williams)

• Frontier AI auditing: Toward rigorous third-party assessment of safety and security practices at leading AI companies (incl. Noemi Dreksler, and Aidan Homewood)

• Third-party compliance reviews for frontier AI safety frameworks (by Aidan Homewood)