NEWS! US and China Launch Superintelligence Dialogue Channel Ahead of Summit

Speaking in New York ahead of upcoming diplomatic talks in Shenzhen, U.S. Treasury Secretary Scott Bessent confirmed that the United States and China have established a bilateral dialogue channel dedicated to Superintelligence (SI) and AI safety oversight.

The initiative establishes formal protocols for exchanging evaluation practices before releasing frontier models, alongside a direct emergency notification mechanism for severe technical incidents or rogue model behavior.

• Emergency Incident Hotline. Both superpowers agreed on a direct government-to-government channel to report critical AI safety incidents, severe anomalies, or unexpected capability leaps before they trigger national security miscalculations.
• Pre-Release Evaluation Sharing. The framework encourages exchanging testing protocols for advanced models and maintaining human oversight over critical systems.
• Competition Remains Unchanged. Secretary Bessent explicitly clarified that cooperation on shared catastrophe risks does not erase underlying rivalries: “Differences won’t vanish through cooperation.” Neither nation committed to capping technical progress or easing advanced semiconductor export controls.

In frontier AI governance, establishing direct bilateral communications mirrors Cold War-era nuclear hotlines. As models approach human and post-human capability levels, the risk of uncoordinated deployment, mutual distrust, or misinterpreting a lab incident as a state-sponsored attack increases exponentially.

By separating shared existential risk management from technological competition, both nations attempt to mitigate acute catastrophe vectors without requiring a freeze on underlying research.