AI Policy Wiki
Dashboard
⚲
AI Policy Wiki › People & Organizations

Scott Bessent

high confidence · updated 2026-09-26

U.S. Treasury Secretary in the Trump administration. Took over the White House AI policy portfolio (with Susie Wiles) in March 2026 after David Sacks's departure. Played a direct role in the April 17 Mythos meeting with Dario Amodei and the April 16 banks-summit on Mythos cybersecurity exposure, and leads the U.S. side in the 2026 AI talks with China.

Contents

Scott Bessent is the U.S. Secretary of the Treasury in the second Trump administration (January 2025–present). In March 2026 he assumed the White House AI policy portfolio jointly with Susie Wiles following the departure of David Sacks as White House AI czar. He leads the American side in the 2026 AI talks with China. His AI-related activity has otherwise centered on the financial-sector dimension of AI policy, including a meeting with Dario Amodei on April 17, 2026 and a summit with major U.S. banks on April 16, 2026 over the cybersecurity exposure associated with Anthropic's Mythos cyber-capability case.

Role in U.S. AI policy

According to May 4, 2026 New York Times reporting (Mickle, Barnes, Frenkel, Volz), Bessent and Wiles stepped in to fill David Sacks's role when Sacks left the White House AI position in March 2026, and have told people outside the administration that they plan to have a larger hand in shaping AI policy (Source: nytimes.com). The same reporting positions Bessent as a business-pragmatist counterweight to the more ideological wing of the administration, contrasting him with Sacks's departure and with Defense Secretary Hegseth's "ideological lunatic" comment about Amodei.

Bessent and Wiles are reported to be among the architects of a proposed working-group executive order on pre-release vetting.

A Wall Street Journal account published September 16, 2026 describes his part in that order's fate. After David Sacks got Trump in May 2026 to scrap the planned order, which would have subjected AI models to an extended government review, Bessent and Wiles were caught off guard and unhappy; Bessent worried that AI-led cyberattacks could wreak havoc on the nation's banking system, according to people familiar with the matter. At the urging of Wiles and Bessent, Trump later signed a slimmed-down version (EO — Promoting Advanced AI Innovation and Security (Trump, signed June 2, 2026)). Only the opening of the article was retrievable (Source: wsj.com). The account is consistent with Lawfare's May 2026 report that Bessent still wanted the order signed after its postponement.

His role extended to the vetting-body design itself. Bloomberg reported on July 17, 2026 that the administration was considering a FINRA-like watchdog to vet top AI models, developed with Bessent, which would report to the SEC and was under review by Wiles — the governmental counterpart to the industry-funded Standards Body proposed three days earlier by Demis Hassabis (A Framework for Frontier AI and the Dawning of a New Age (Hassabis, July 2026); bloomberg.com). The SEC reporting line places the proposal within Bessent's own portfolio rather than Commerce's.

Axios reported in September 2026 that Bessent was among the few cabinet members who flanked President Trump at a G7 press conference in June, at which AI chief executives and the administration discussed a U.S.-led effort to coordinate global AI standards, that he helped cool Anthropic's relations with the administration after its dispute with the Pentagon, and that sources familiar had previously described him as a reasonable actor trusted by the private sector and critical-infrastructure operators (Source: axios.com).

Response to the Mythos cyber-capability case

Bessent's first sustained AI engagement came through the federal response to Anthropic's Mythos cyber-capability case in April 2026. On April 16 he convened the largest U.S. banks for talks about Mythos's potential cyber-capabilities, which were then being framed as having the potential to "lead to a cybersecurity 'reckoning'" (Source: Economist, April 16, 2026) — an early instance of Treasury involvement in the response.

The following day Bessent and Wiles met with Dario Amodei at the White House, focused on returning Anthropic's technology to government use following the Pentagon–Anthropic supply-chain-risk dispute. Both sides later described the meeting as "productive" (Source: nytimes.com). The meeting preceded the May 2026 administration shift toward pre-release vetting and the announced CAISI pre-release-evaluation expansion.

U.S.–China AI talks

Bessent has been the administration's principal channel for AI discussions with China. At the Trump–Xi summit in Beijing he said on May 14, 2026 that the United States and China would "set up a protocol" on best practices to keep dangerous AI capabilities out of nonstate actors' hands, a remark described as the summit's only concrete AI deliverable (Source: insideaipolicy.com). Sen. Jim Banks (R-IN), in a letter released by his office on August 6, 2026, urged Bessent to determine whether the two countries could work together to mitigate AI risks, writing that undisclosed models "have important implications for your upcoming engagements with the PRC on advanced AI" (Source: insideaipolicy.com). Two sources briefed on the planned mid-September talks told Reuters on September 4, 2026 that Bessent would lead the American side, and that Washington wanted to discuss joint monitoring of AI-directed cyberattacks, had floated asking U.S. and Chinese labs to "police themselves" and share information, and intended to raise alleged Chinese distillation of U.S. models (Source: reuters.com).

On September 16, 2026, Bessent said in a statement to Axios: "The United States remains the leader in AI. And we are open to discussions on avoiding shared risks and avoiding bifurcation of our two systems. We expect our discussions to cover both open and closed weight models." He was scheduled to meet Chinese Vice Premier He Lifeng that weekend in multi-hour meetings in New York City covering AI, trade, rare earths and other economic issues, with possible working-level meetings continuing into the following Monday, ahead of the Trump–Xi summit in Washington on September 24. A senior administration official told Axios that "Bessent and [United States Trade Representative Jamieson Greer] are the point guys on China," while Commerce Secretary Howard Lutnick had led AI discussions at the G20 summit in North Carolina earlier that month (Source: axios.com). The statement came the day after his House Financial Services Committee testimony calling for more U.S.-built open-source models to "push back against China" (below). See US-China AI Competition: Different Races, Different Metrics.

After talks with Vice Premier He Lifeng in New York on September 20, 2026, Bessent said the United States had proposed to China a "notification mechanism" for AI incidents rising to the level of national security concern (Source: axios.com). On CNBC the next day he called it "a communications line, an incident line," said the two countries had formalized the "USA-China AI dialogues," and said senior officials would meet again in about two months in Shenzhen; he also said OpenAI's Hugging Face incident "is the responsibility of the OpenAI management, not a bunch of agents," and that labs "can slow down anytime they want to" (Source: reuters.com). Four people familiar with the arrangement described the mechanism as an open line of communication between Bessent and He rather than a formal body of experts, and said Chinese officials had not publicly agreed to it (Source: politico.com). Administration officials said on September 23 that the notification system would take a number of weeks to finalize, targeting a framework agreed to by Trump by the time Bessent and He meet again in Shenzhen, with benchmarks for which incidents qualify and explicit Chinese sign-on still unresolved (Source: wired.com).

Bessent emerged as a frontrunner for the AI czar post Trump announced on September 19, 2026, according to three people familiar with the matter in an account that became public on September 22; OSTP Director Michael Kratsios and Office of Personnel Management Director Scott Kupor were also under consideration, and Trump's decision was not final (Source: semafor.com). A separate account on September 23, citing two sources, named Bessent, Commerce Secretary Howard Lutnick, Kratsios and Sean Cairncross as in contention, and said some officials expected the incident-notification system to wait on the choice (Source: wired.com). The two accounts' lists of candidates differ. On September 25, 2026 Trump wrote on Truth Social that "Scott Bessent will not be going to be Super Intelligence (SI) Czar," adding, "Number One, he doesn't want to. Number Two, he's doing such a great job at Treasury, and that's where I want to keep him!" (Source: cnbc.com).

Treasury-side AI engagement

Bessent's Treasury portfolio intersects with AI policy along several lines. Treasury's cybersecurity-protection mandate, exercised through bodies such as the FFIEC and OCC, is being reframed around frontier-AI offensive-cyber risk. The Treasury Secretary's authority also touches agentic-finance applications, including Anthropic's May 5, 2026 financial-services agents and the FIS Financial Crimes AI Agent launch. Treasury's regulatory posture on OpenAI's planned IPO sits within the broader White House recalibration; per Brockman trial testimony, OpenAI's 2026 compute spend was stated at $50 billion.

Public positioning

Bessent's public AI commentary has been more restrained than Sacks's. During the Mythos period he declined to publicly defend or attack Anthropic, framing the Treasury portfolio in risk-management terms.

On July 21, 2026, Bessent said the US supports open-source AI but "we have the ability to sanction" overseas models that steal from US companies via distillation — a Treasury-enforcement framing of the distillation dispute that had previously been argued mainly in export-control and legislative terms — citing "watermarks" of U.S. models found inside Chinese ones (Source: cnbc.com). In a follow-up post on X he warned that "sanctions and Entity List designations will be on the table" for "covert, industrial-scale distillation attacks" (Source: x.com). See Adversarial Distillation.

At a House Financial Services Committee hearing on September 15, 2026, asked by Rep. Juan Vargas (D-California) what the administration was doing on AI safety, Bessent said Treasury had been "working on safety nonstop since the release of Mythos," and that what the government "shouldn't do on safety is to give these labs a liability exemption, which is what they are asking for." He continued: "The best way to guarantee safety is that the creators are liable for what they build and generate," and "they are saying that 'we would like to all slow down, but please give us a waiver on liability,' which should not be done. And I would encourage everyone in this committee and in both houses not to consider that." He told Rep. Ritchie Torres (D-New York) that after the July Hugging Face incident the administration had met with AI laboratories and the financial sector, told Rep. Josh Gottheimer (D-New Jersey) that Treasury was extending lessons from the financial sector to other sectors through the Gold Eagle forum, a vulnerability clearinghouse it runs with CISA, and called for more US-built open-source models to "push back against China": "We can't let these large labs have regulatory capture because that will stop innovation" (Source: fedscoop.com). The waiver requested in Amodei's pacing essay was an antitrust waiver for safety conversations rather than a liability exemption; the same account notes OpenAI's earlier support for an Illinois bill limiting liability for AI-caused harms. See AI Liability, Coordinated slowdown proposals compared.

Relationships

Sources

  • "White House Considers Vetting A.I. Models Before They Are Released" — NYT, May 4, 2026 (Source: Raw Sources/White House Considers Vetting A.I. Models Before They Are Released.md).
  • (Source: nytimes.com) — April 17 Wiles+Bessent+Amodei meeting.
  • (Source: Economist, April 16, 2026) — Bessent summons banks on Mythos.
  • (Source: axios.com) — September 16, 2026 statement on AI talks with China.
  • (Source: reuters.com) — Bessent to lead the U.S. side in the September talks.
Categories: People & Organizations | us | china | government | person | treasury | ai-policy | national-security
What links here (38)