Google DeepMind chief Demis Hassabis calls for U.S. to spearhead AI standards body

The leader of Google's AI division has urged the United States to take the lead in establishing a governing body for artificial intelligence that would evaluate new models and address national security concerns, including cybersecurity and biological threats. Demis Hassabis, head of Google DeepMind and a Nobel laureate, expressed the need for immediate action to manage risks associated with artificial general intelligence (AGI)—the stage at which AI equals or exceeds human intelligence. In a post on X, Hassabis highlighted that the challenges posed by frontier AI models have already impacted cybersecurity, with potential future risks from nuclear and biological threats as capabilities advance. He proposed a collaborative initiative between the public and private sectors, under federal oversight, as a means to effectively confront these dangers. This call for action follows recent discussions among tech leaders, including a meeting at the G7 where Hassabis and Anthropic CEO Dario Amodei advocated for a U.S.-led coalition to create regulatory frameworks for AI. OpenAI's Sam Altman also echoed similar sentiments in a recent Financial Times article, illustrating a growing consensus among industry leaders for an AI oversight body. Despite these calls for regulation, the debate over how to govern leading AI technologies has become increasingly contentious between the public and private sectors. Anthropic has recently engaged in negotiations with officials after the Trump administration imposed temporary export controls on an advanced AI model, while OpenAI faced its own restrictions regarding the rollout of a new model. Hassabis emphasized that the U.S. is ideally positioned to lead the development of an AI regulatory framework, given its economic and technical capabilities. He suggested the establishment of a new standards body, similar to the Financial Industry Regulatory Authority (FINRA), which would include independent technical experts and representatives from open-source communities. To ensure the proposed body could function effectively, Hassabis indicated that substantial funding would be required, likely sourced from the industry. Initially, leading AI research labs would voluntarily share their models with this body for review, with mandatory submissions anticipated for U.S. market deployment once models are proven effective. He also mentioned the potential for specific tests to evaluate AI systems, focusing on safety measures and ethical practices, such as digitally watermarking AI-generated content and generating human-readable output tokens to clarify model reasoning. With the ongoing competition between the U.S. and China in AI development intensifying, the urgency for regulatory oversight is accentuated. Recent releases from Chinese companies, such as DeepSeek and Z.ai, have raised concerns among U.S. lawmakers about the growing influence of foreign AI technologies in the domestic market, prompting considerations on how to manage the adoption of these models by American companies.