Partnering with Microsoft to strengthen frontier AI safety

Today we’re announcing a new partnership with Microsoft that will focus on developing and advancing methods for evaluating high-risk AI capabilities, and the effectiveness of safeguards used to address them. The partnership will also include research into societal resilience, examining how conversational AI systems interact with users in sensitive contexts. As AI systems become increasingly capable, sustained two-way collaboration between government and companies developing and deploying frontier AI is essential to advance our joint understanding of large-scale risks to public safety and national security.