President Donald Trump and six leading AI company executives signed a voluntary safety accord at the White House on Tuesday, September 29, 2026, calling for internal safety controls and independent audits — a pledge that carries no legal enforcement mechanism, drawing immediate criticism from lawmakers and AI policy experts.
Who Signed the Accord
The signatories were Dario Amodei (Anthropic), Greg Brockman (OpenAI), Sundar Pichai (Google), Mark Zuckerberg (Meta), Elon Musk (xAI), and Jensen Huang (Nvidia), according to reporting from Al Jazeera and other outlets. Trump posted the one-page document on Truth Social following the signing.
What the Accord Actually Asks For
The agreement calls for four layers of AI controls and audits: robust internal safety controls, independent outside auditors to test whether those controls actually work, a board-level committee at each company to review audit findings, and a commitment that "participating companies will meet regularly to establish standards and best practices to improve the safety of their systems." The text itself acknowledges that formal legislation "may make sense" at some future point, without committing to it.
A Separate, Symbolic Move
Alongside the accord, Trump signed an executive order renaming artificial intelligence to "Super Intelligence," or "SI," requiring all federal agencies to use that terminology going forward. The order does not establish any new rules or requirements for the AI industry — leaving the voluntary accord as the primary safety commitment currently on the table.
No Enforcement Mechanism
Despite the accord's stated goals, it cannot be legally enforced. Trump described the agreement as "morally binding" rather than legally binding, and told reporters he was "seeing tremendous self-policing" across the industry already. Asked about the lack of enforcement, Trump argued: "We automatically have regulation with the Department of Justice, the FBI, all of that," according to Al Jazeera. He said he would name an overseer in the coming days and suggested a committee of roughly 10 people would monitor the accord's implementation.
A Nearly Identical Precedent From 2023
This isn't the first time most of these companies have made this kind of promise. Five of the six companies represented at Tuesday's signing had already made a similar voluntary AI safety pledge to the previous administration in July 2023, covering testing, security, and watermarking commitments. That earlier pledge also ended with a promise that the administration would "pursue bipartisan legislation" — a promise that, three years later, had not resulted in any such law. Critics at the time also flagged vague language in the 2023 pledge, noting it didn't specify who the "independent experts" conducting safety testing would actually be, or what testing they would perform.
Senator Blumenthal's Criticism
Senator Richard Blumenthal (D-Conn.) sharply criticized the pledge during a Senate hearing the day after the signing, calling it "worse than ineffectual." Lawmakers in both chambers of Congress have introduced several AI-related bills in recent weeks, reflecting continued legislative pressure that this voluntary accord does not resolve.
The Key Distinction Critics Are Raising
Policy observers have drawn a pointed distinction between the kind of after-the-fact auditing this accord describes and a genuine pre-release safety regime — a system where an independent party has the actual authority to say no before a model ships, rather than simply reviewing and reporting on safety controls after the fact. The current accord does not establish any such pre-release approval authority.
Context: Why This Accord Emerged Now
The agreement arrives amid a notably active stretch of AI safety discourse, following public calls earlier in September from Amodei, OpenAI's Sam Altman, and Musk urging the industry to slow development of its most advanced models, and amid broader scrutiny following reported incidents of AI agents acting outside their intended operating constraints. It also follows Anthropic's own IPO prospectus, which separately disclosed warnings about "catastrophic or existential risks" tied to its AI models — underscoring that even companies signing onto a voluntary industry accord continue flagging serious risk concerns through other channels simultaneously.
What's Next
With Trump still expected to name an overseer and finalize the oversight committee's composition, and Congress continuing to weigh its own AI legislation separately, the coming weeks will reveal whether this accord functions as a meaningful first step toward binding regulation, or repeats the pattern of its 2023 predecessor, whose promised follow-up legislation never materialized. For continuing coverage, see the full report from Technology.org.