The notion that tough export controls deny diplomatic space for AI risk discussions with China is a "mental model error." The Biden administration proved it's possible to compete vigorously by implementing chip restrictions while simultaneously engaging in government-to-government dialogue on AI-enabled nuclear risk.

The most realistic path for US-China AI talks involves creating protocols to prevent models falling into the hands of terrorists or criminal groups. Cooperation on preventing state-level cyberattacks is unlikely, as both sides assume the other's intelligence agencies will use AI for espionage.

Tech leaders state they would support an AI development pause if competitors, especially China, also agreed. This is a strategic PR move, as they know a global consensus is unachievable. It allows them to appear responsible about AI safety without any actual risk of having to slow down progress.

Stopping frontier AI development in the US without a verifiable, lockstep agreement with China does not increase safety. Instead, it creates a critical vulnerability, ceding dominance in cyber warfare and national security to an adversary who will continue to advance their own AI capabilities.

Anthropic CEO Dario Amadei’s call to slow AI development is explicitly conditional on maintaining a lead over China. This framing, which includes calls for stricter chip export controls, turns a global safety issue into a geopolitical one, prompting China to label it a "silent AI cold war" and reject cooperation.

The unstated reason AI labs are appealing to the US government to "pace" development is that only a government can handle the necessary international diplomacy. The labs recognize that any meaningful slowdown is futile without China's participation, making the letter an implicit request for the US to initiate a global, coordinated agreement.

Global consensus on AI safety is unlikely soon. The most practical approach is a bilateral agreement between the US and China, the two dominant players. This can begin with informal 'track two' conversations between their respective AI labs before escalating to a formal government pact.

A pragmatic starting point for U.S.-China AI cooperation is to agree on verifiable red lines for proliferating dangerous dual-use capabilities, such as advanced cyberattack tools. This addresses a mutual security interest and builds the institutional trust and processes needed for more ambitious agreements on superintelligence.

Chinese AI firms maintain close, ongoing communication with the Cyberspace Administration of China (CAC), including informal weekly meetings. This relationship ensures regulators are never caught by surprise, making it virtually impossible for a company to release a powerful new model without prior notice.