President Donald Trump and six leading technology executives signed a voluntary AI safety accord at the White House on Tuesday, 29 September, setting out four layers of controls and audits for companies developing and deploying powerful frontier models.
The agreement does not establish federal penalties or an enforcement mechanism. Trump nevertheless described it as 'morally binding' and compared it to a constitution, while arguing that the technology industry was already showing 'tremendous self-policing'.
'It's almost like a Constitution in a way, and the biggest people in the world signed that and I signed it as president and it really is a form of protection,' Trump told reporters after meeting technology leaders in Washington.
The published document is titled the White House Accord on Super Intelligence.
House Speaker Mike Johnson described it as a voluntary 'statement of principles' setting expectations for robust internal controls and layers of internal and external review.
What the AI Accord Calls For
The accord sets out what it describes as 'four layers of controls and audits' intended to identify and address problems while frontier models are being trained and after they are deployed.
The first calls on participating companies to implement robust internal controls monitoring the capabilities and alignment of their models during training and deployment. The document specifically names cybersecurity, biosecurity and chemical threats, while calling for controls intended to ensure models do not hack or access technical systems in unintended ways.
The second layer calls for an internal team to check whether those controls, monitoring systems and detection measures are working as intended and whether identified problems are addressed.
Companies would then work with an independent external auditor or evaluator to assess whether their controls, monitoring and detection systems are operating properly.
Finally, each company would designate an independent committee of its board of directors to receive reports from the teams running the controls and from internal and external auditors or evaluators. The committee would also oversee efforts to remedy identified problems.
The published accord does not specify how the independence of outside evaluators or board committees would be determined. Nor does it require audit findings to be made public. The document instead leaves implementation largely to the participating companies.
Those companies also committed to meeting regularly to establish standards and best practices aimed at improving the safety of their systems.
The accord was signed by Google CEO Sundar Pichai, Anthropic CEO Dario Amodei, Meta CEO Mark Zuckerberg, OpenAI President Greg Brockman, Nvidia CEO Jensen Huang and xAI founder Elon Musk, along with Trump.
Why the Pact Is Voluntary
The accord does not create a new federal enforcement system or specify legal penalties for companies that fail to meet its commitments.
Instead, its four-layer framework places responsibility on the companies themselves, their internal oversight teams, external auditors or evaluators and corporate boards.
Asked whether the agreement was binding, Trump said: 'I think it's morally binding.'
He later told reporters: 'I think I'm seeing tremendous self-policing, and they understand that they have to self-police.'
The arrangement could eventually change.
The accord states that 'over time' it may make sense to codify the measures into laws or regulations. It adds that the participating companies regard the controls and audits as important regardless of whether they eventually become compulsory.
Johnson described the agreement as a 'statement of principles' containing voluntary industry commitments and 'robust internal controls and layers of internal and external review'. He said Congress would continue assessing the rapidly developing technology.
How Trump Framed AI Self-Regulation
Trump presented the accord alongside his broader argument that the US should allow rapid AI development while relying heavily on companies to police their own systems.
Earlier on Tuesday, Trump said he would 'never stifle the growth' of technology that he said could become bigger than the Industrial Revolution. He also portrayed advanced AI as a global competition and said the US was 'leading by a lot', including over China.
Asked separately about regulation, Trump said there was a belief that there should be 'tremendous self-regulation'. He added that government oversight already existed through institutions including the Department of Justice and FBI.
The accord seeks to address risks surrounding increasingly powerful frontier models through internal monitoring, outside assessments and board-level oversight, without turning those commitments into mandatory federal standards.
Trump also signed a separate executive order titled Inaugurating the Era of Super Intelligence. It directs executive departments and agencies, 'to the maximum extent permitted by law', to use the terms 'Super Intelligence' and 'SI' instead of 'Artificial Intelligence' and 'AI' in official correspondence, public communications, websites, reports, policy documents and other non-statutory documents.
The order does not require previously issued regulations, presidential actions, contracts, grants or historical documents to be rewritten. For now, it uses the existing statutory definition of artificial intelligence to determine what 'Super Intelligence' means, unless Congress or a later presidential action changes it.
It also directs the Assistant to the President for Science and Technology to submit proposed legislative language for a federal definition of 'Super Intelligence' and 'SI' within 60 days.
That terminology is reflected in the name of the new White House accord.
For now, the agreement gives participating AI companies a common voluntary framework centred on internal controls, outside assessment and board-level oversight. Whether those commitments eventually become enforceable US rules remains unresolved, although the accord explicitly leaves that possibility open.
