Scoop: U.S. open to discuss AI "shared risks" with China, Bessent says

TEXT START: The U.S. is open to discussing shared risks with China in upcoming AI talks this weekend, Treasury Secretary Scott Bessent told Axios.

The Dissection

This is diplomatic risk-management theater wrapped around an arms race. The United States is willing to discuss AI danger while explicitly refusing to slow AI development. The message is not “we will restrain the machine.” It is “we will compare notes while accelerating it.”

The article frames communication as governance. It is not. It is an attempt to reduce strategic miscalculation between rival states without touching the underlying incentive: whoever captures superior AI gains economic and military leverage.

The Core Fallacy

The text treats dialogue as if it can overcome competitive compulsion. Under the Discontinuity Thesis, that is the central error. P1 makes advanced AI economically dominant; P2 prevents institutions from reliably preserving human-only domains at scale; P3 destroys mass productive participation. Talks about “shared risks” do not alter those mechanics.

A safety conversation that leaves the race intact is not a brake. It is a maintenance inspection on a vehicle whose owners have agreed not to stop driving.

Hidden Assumptions

• That U.S. leadership in AI is durable rather than a temporary advantage requiring continuous escalation.
• That China and the U.S. can distinguish “shared risks” from the strategic advantages they are unwilling to surrender.
• That states can coordinate tightly enough to constrain AI while still competing for dominance.
• That discussing catastrophic outcomes produces enforceable limits rather than better methods of managing escalation.
• That economic disruption can be handled after the fact through policy without confronting the collapse of labor’s productive role.
• That the social order survives merely because governments remain capable of issuing statements and transfers.

Social Function

Transition management, prestige signaling, and elite self-exoneration. The article gives decision-makers the appearance of responsibility while preserving the policy that drives the discontinuity. It converts “we are accelerating a system-breaking technology” into the more palatable “we are responsibly discussing shared risks.”

The Verdict

This is not evidence that the post-WWII order can regulate its successor. It is evidence that its governing class recognizes the danger but remains structurally unable—or unwilling—to sacrifice competitive advantage to prevent it. The talks may reduce diplomatic friction. They do not reverse AI dominance, restore mass productive participation, or save the wage-consumption circuit. They are a protocol for managing the approach of obsolescence, not a mechanism for stopping it.