Product1 publisher3 min readPublished
White House shelves the self-regulating AI body Treasury and OSTP had drafted
Wired reports the FINRA-style oversight framework has sat in limbo since executives called Trump in August. Product teams are left with an embedded-auditor bill and no House voting days before the midterms.
The Product Desk · Product desk
What happened
- The bipartisan Frontier Act would let the Commerce Department embed auditors at AI labs to report critical safety incidents and let the administration stop development of a catastrophically risky model.
- The House has no voting days left until after the midterm elections and no unity behind a single AI bill. Wired reports the odds of passage are extraordinarily slim.
- Two House leadership aides said Democrats are looking at attaching some of the bill's audit provisions to must-pass legislation such as a continuing resolution.
Compiled by The Product DeskSomething wrong?How this is made
Why it matters
- decision Teams that parked their frontier-model documentation until a federal template arrived now have to pick their own standard and defend it in customer security reviews, with no agency framework to point at.
- constraint If audit language reaches law by riding a funding bill, the lead time is whatever a shutdown deadline allows. There is no rulemaking notice or comment period to prepare against.
- exposure The surviving federal option puts government staff inside the lab reading incident reports, so internal eval and escalation records become documents an outside party may read.
- contradiction Executive phone calls killed the lighter self-regulation draft while OpenAI publicly endorses the heavier auditor bill, so a vendor saying it supports AI regulation tells a buyer little about which rules it will accept.
Plenty of frontier-model plans carry a line about aligning with a forthcoming federal framework. Wired reports that the framework in question was a preliminary draft, written at the Treasury Department and the White House Office of Science and Technology Policy, for a body in which leading AI labs would self-regulate the development of frontier models [3]. It has been in limbo since a run of August calls to President Trump from tech executives including David Sacks and Mark Zuckerberg [4]. Officials paused the work after Trump soured on the idea [1].
The proposal came from outside government. Demis Hassabis floated it in a July 14 essay and in briefings to senior officials, and White House aides began sketching soon after, according to sources Wired described as familiar with the matter [2]. What cooled the building further, per those sources, was a 3,822-word warning from Anthropic CEO Dario Amodei about the dangers of frontier AI, which aides read as stirring anxieties and causing headaches for the administration [7]. Trump has since called AI safety concerns a "hoax" in a Truth Social post [6].
A self-regulating body run by the labs would have produced a standard a vendor could cite in a security review. The Frontier Act, introduced this summer by Lori Trahan and Jay Obernolte and reported as having the most support in the House Democratic caucus, does something different [9]. As drafted it would let the Commerce Department embed auditors at AI labs who report critical safety incidents, and it would let the administration stop development of a model that poses catastrophic risk [10].
Wired's account puts several obstacles between that bill and a floor vote [16]. The House has no voting days left until after the midterm elections, and no single bill has unity behind it [8]. Speaker Mike Johnson has no interest in bringing it up, House Democratic leadership aides told Wired [12]. Support is incomplete among Democrats too, with Josh Gottheimer, who sat on the House Democrats' AI commission, not signed on [13]. Two House leadership aides said Democrats are looking at putting some of the audit provisions into must-pass legislation such as a continuing resolution [14].
Support for regulation means several different things here. OpenAI publicly backs the auditor bill while the White House remains undecided, facing objections from accelerationists about regulatory capture and from national security hawks about China [11]. The kill-switch bills have bipartisan and bicameral support and only modest interest from Silicon Valley executives [15].
This reporting covers the White House and Congress, and it does not address state legislation or contract terms [17]. Your next enterprise security questionnaire is outside its scope. What it establishes is that no federal artifact is arriving on a published calendar this session.
Two questions sort the safety work on a roadmap: whether an outside party gets read access to the artifact, and whether a signed agreement requires the artifact to exist. Answer yes to both and you have a commitment someone can hold you to. In this reporting the only federal candidate for the read-access box is the Frontier Act's audit language, and the Speaker will not schedule it [10][12].
What to watch
- Whether Frontier Act audit provisions appear in the text of a continuing resolution or another must-pass vehicle.
- Whether the White House takes a position on independent auditors after being reported as undecided.
- Whether any new oversight proposal reaches serious consideration at the White House after the paused FINRA-style draft.