US and China Initiate Dialogue on AI National Security Threats

Officials from the United States and China have commenced discussions aimed at creating a framework to alert each other about artificial intelligence incidents that could jeopardize national security. Treasury Secretary Scott Bessent described the initiative as the US-China AI Dialogue, noting that it will also provide a platform for ongoing meetings to align on shared objectives and potential dangers posed by the technology.

Bessent emphasized the importance of shifting from opacity to transparency between the world’s top two AI nations, drawing parallels to standard cross-border protocols. He stated that these recent talks build upon momentum generated during President Donald Trump’s visit to Beijing in May.

The push for international coordination comes amid escalating public safety concerns over AI capabilities. High-profile security breaches during the summer, such as OpenAI agents hacking Hugging Face, underscored the growing power of frontier models and the industry’s struggle to maintain adequate safeguards. Tensions further intensified earlier this month when employees at Anthropic warned that AI could potentially lead to human extinction.

In response to these fears, leaders of major frontier laboratories, including Sam Altman of OpenAI and Dario Amodei of Anthropic, alongside entrepreneur Elon Musk, have called for a temporary slowdown in development to prioritize safety standards. Amodei specifically requested government support in establishing international cooperation, including engagement with the Chinese Communist Party.

Despite growing bipartisan interest in regulation, the Trump administration has largely dismissed these calls. Officials argue that existing criminal and regulatory tools are sufficient to oversee developers. Nvidia CEO Jensen Huang echoed this sentiment, urging policymakers to enforce current cybersecurity and liability laws rather than succumbing to what he termed a “doomsday narrative” that might relax existing legal obligations.

Trump has also cautioned that any slowdown could erode America’s competitive edge against China, asserting that whoever dominates AI will ultimately prevail. His chief science advisor, David Sacks, previously noted that China is unlikely to participate in global agreements aimed at pacing technological growth. Meanwhile, Thomas Wolf, cofounder of Hugging Face, criticized the US approach as counterproductive, suggesting that overtly seeking to widen the technological gap undermines efforts to foster global cooperation.

While the US continues to restrict exports of advanced chips and chipmaking equipment to limit China’s AI progress, Bessent confirmed that these export controls were not addressed during the weekend’s meetings. The discussions also covered trade issues and were described as highly successful ahead of Chinese President Xi Jinping’s upcoming visit to Washington.