US Treasury Secretary Bessent has put forward a proposal to build a US-China artificial intelligence (AI) safety system, a claim that held up under a source review completed by this desk. The verification, based on 11 primary sources and archived in a report dated September 21, 2026, found that the mechanism exists only at the proposal stage — it has not been built, and China's position remains unconfirmed — a status that does not contradict the claim as worded.
The verified core of the claim is narrow and specific: Bessent proposed constructing an AI safety system between the United States and China. Reviewers noted that no such system has been established and that Beijing has not finalized or confirmed any commitment to it. Because the claim describes a proposal to build the system — not its completion or operation — the gap between announcement and implementation does not clash with the claim's wording.

출처: 조사 출처
The desk examined 11 primary sources in the course of the review and stored its findings at reports/2026-09-21_bessent_us_china_ai_safety_mechanism.md. Imagery associated with the story — an official portrait published by the US Department of the Treasury and photographs distributed through CNBC and Reuters — was preserved alongside the file. One Reuters link tied to the review could not be opened at the time of writing and has been flagged as blocked, pending direct capture of the page.
Bilateral coordination between Washington and Beijing on AI safety is followed closely by Korea's AI and semiconductor community, where companies and policymakers track any framework that could shape norms for advanced computing and large model development. For now, the record shows a proposal from the US side with no confirmed Chinese commitment — nothing more, and nothing less. On the strength of the source record — the proposal itself, its unexecuted status, and China's unconfirmed stance — this desk's final determination on the claim is True.