Gates Pushes China, US for AI Bioweapon Safeguards

Bill Gates is working to arrange a meeting with Chinese President Xi Jinping later this year to discuss coordinated international measures on artificial intelligence risks. The Microsoft co-founder told Reuters he plans to urge nations to monitor AI systems for their ability to synthesize molecules that could enable biological attacks. He expressed hope that if the United States leads on restricting release of certain hazardous models, other countries including China would follow.

Staff for Gates confirmed efforts to schedule the session during a planned November visit to China. Gates noted he met Xi three years earlier as one of the first foreign business figures hosted by the Chinese leader after the pandemic. In comments to reporters, he suggested China has already acted to limit children's exposure to addictive AI companions and might respond to broader proposals on safety.

Gates has shifted from earlier statements that viewed AI as largely beneficial. In a recent essay exceeding 5,000 words he described the technology as creating one of the most turbulent periods in history and warned that comparisons to past inventions like personal computers understate the speed and scope of change. He cited potential for cyberattacks, workforce displacement and user dependence as areas requiring attention.

The push for monitoring comes as Gates argues this technological shift differs from previous ones because of its rapid capability gains. He has said he would likely back efforts to slow global AI progress if a credible plan existed, though he maintains that safeguards on molecule synthesis need not halt commercial development.

Historical patterns in technology adoption show markets often address emerging hazards through competition and liability rules rather than multilateral pacts. Attempts at uniform international standards have frequently produced uneven compliance, with leading firms gaining advantages from compliance costs that smaller competitors struggle to meet. Gates' own background in software reflects periods of rapid innovation under relatively light early oversight, when product improvements outpaced regulatory frameworks.

Chinese authorities have demonstrated willingness to impose controls on technology use within their borders, yet enforcement of shared global rules on advanced models would require verification mechanisms that sovereign states have resisted in other domains. Proposals for monitoring AI outputs tied to biological risks also raise questions about how such systems would distinguish legitimate research from prohibited activity without creating broad surveillance structures.

Data from prior waves of automation indicate that labor market adjustments occur over time through new job creation rather than permanent displacement, though transition periods involve real costs for affected workers. Gates' emphasis on policy management by society as a whole aligns with calls for centralized direction that have produced mixed records when applied to complex technical fields.

Whether the proposed meeting yields concrete agreements remains uncertain. Past technology conferences have generated declarations on cooperation without binding commitments that alter competitive incentives among nations or companies.