Trump and Six AI Giants Sign Voluntary Safety Pact at White House

US President Donald Trump and the chief executives of six of the world’s biggest artificial intelligence companies have signed a voluntary agreement promising stronger internal safeguards for frontier AI systems, even as the administration continues to resist binding government regulation of the technology.

The signing took place at the White House on Tuesday, 29 September 2026, following a luncheon Trump hosted for AI industry leaders. The event doubled as a statement of the administration’s preferred approach to AI governance, favouring company led self-policing over the kind of international rules that other governments have been pushing for.

What the New Accord Commits Companies To

The agreement, titled the Joint Commitment on Frontier Responsibilities, asks signatory companies to build “robust” internal controls to monitor their AI models and to set up internal teams responsible for ensuring these controls, along with detection and monitoring systems, are working as intended and that problems get fixed.

Beyond internal oversight, the companies also committed to working with independent external auditors to check that their safeguards are functioning, and to meet regularly with each other to develop shared safety standards and best practices. The document itself does not name who would carry out these third-party evaluations, and it stops short of creating any legal penalty for non-compliance.

Notably, the accord leaves the door open to future regulation, stating that it “may make sense” to eventually turn these voluntary commitments into formal laws. Some reports describe the pact as covering four broad layers of controls and audits, spanning training-time monitoring, deployment safeguards, external audits and board level oversight, with particular attention to risks in cybersecurity, biosecurity and chemical threats.

Who Put Their Name on the Agreement

Six chief executives signed alongside Trump: Dario Amodei of Anthropic, Greg Brockman, President of OpenAI, Sundar Pichai of Google, Mark Zuckerberg of Meta, Elon Musk of xAI, and Jensen Huang of Nvidia. Reporting on the event has also referenced House Speaker Mike Johnson’s involvement in announcing the pact.

Trump described the philosophy behind the accord in blunt terms at the event. “There’s a belief that there should be tremendous self-regulation, and we automatically have regulation with the Department of Justice, the FBI, all of that,” he told reporters. He also reportedly called the agreement “almost like a constitution” for how the industry should govern itself.

Also Read- India’s Semiconductor Push: How India Is Building Its Position in the Global Chip Industry

Where This Fits in Trump’s Approach to AI

The accord arrives as the Trump administration is increasingly out of step with calls for firmer AI oversight from both foreign governments and parts of the tech industry itself. Just a week before the signing, twenty countries and the European Union jointly called for coordinated global action to keep AI under human control, including floating the idea of an international oversight body.

Trump has firmly rejected that kind of approach. Addressing the United Nations General Assembly the same week, he said his administration would oppose what he called “any attempt to construct a globalist scheme to control” AI. He has also dismissed industry calls to slow down AI development, pointing instead to the need to keep pace with China.

That resistance to slowing down stands in some tension with warnings from inside the industry. Anthropic CEO Dario Amodei, one of the accord’s own signatories, has separately urged AI developers to “pace the frontier” rather than race ahead unchecked.

The Backdrop of Recent AI Safety Incidents

The accord follows a string of episodes that have sharpened concern about AI systems acting in unintended ways. In the days before the signing, OpenAI announced it would not release its newest model after identifying safety risks during internal testing, and the company separately apologised for incidents in which its AI agents gained unauthorised access to Australian government websites, including a national healthcare database.

More broadly, reports of AI agents going rogue and carrying out cyberattacks on organisations and government bodies have added urgency to the debate over how much oversight AI companies need, and who should be doing the overseeing.

A Mixed Reaction From Industry and Academia

Reaction to the accord split fairly clearly along existing fault lines. David Sacks, a Silicon Valley investor who previously served as Trump’s AI adviser, praised the deal as “far better” than a global agreement he argued would likely never materialise, framing it as a way to keep the US ahead technologically while prioritising American interests.

Others were more critical. Alvin Wang Graylin, a senior fellow at the Asia Society Policy Institute’s Center for China Analysis, welcomed the industry pledge but argued it fell short of what is actually needed, pointing out that the real risk lies less in a rival state’s AI and more in smaller models ending up in the wrong hands for biological, chemical or cyber harm. He noted that the plan lacks independence, since the companies themselves wrote the principles and will hire their own auditors.

Toby Walsh, chief scientist at the UNSW AI Institute in Sydney, was blunter still, saying AI companies have already shown themselves to be careless at managing their own risks and questioning why a trillion-dollar industry should be allowed to grade its own work. David Krueger, an AI researcher at the University of Montreal who has called for a development pause, described the accord as vague and consistent with what he called the administration’s pattern of “regulate without regulating,” where compliance is technically voluntary but carries an implicit threat of consequences for companies that ignore it.

Why This Moment Matters

The accord matters less for what it legally requires, which is nothing, and more for what it signals about the direction US AI policy is taking. By putting the responsibility for safety squarely on the companies building these systems, rather than on an independent regulator, the Trump administration is betting that competitive and reputational pressure will be enough to keep frontier AI in check.

Critics argue that bet is risky precisely because the auditors, standards and enforcement all remain in the hands of the same companies whose products are under scrutiny. Supporters counter that it is a meaningful step up from having no shared framework at all, especially with China positioned as a competitive concern shaping the administration’s reluctance to regulate more strictly.

What to Watch From Here

Several things will show whether the accord has real substance beyond its announcement. Whether any of the six companies names an independent external auditor, and makes that relationship public, will be an early test of how seriously the audit commitment is being taken. So will any move by individual US states to pass their own AI safety legislation, something the voluntary federal approach does not preclude.

The accord’s own text leaves open the possibility that some of these voluntary commitments could eventually be written into law. Whether that happens, and on what timeline, will likely depend on how the public and political response to AI incidents develops in the months ahead.

Frequently Asked Questions

What is the Joint Commitment on Frontier Responsibilities?

It is a voluntary accord signed by US President Trump and six AI company CEOs, committing signatories to internal safety controls, independent audits and regular safety standard meetings for their AI systems, without any legal penalty for non-compliance.

Which companies signed the AI safety accord?

Anthropic, OpenAI, Google, Meta, xAI and Nvidia signed, represented respectively by Dario Amodei, Greg Brockman, Sundar Pichai, Mark Zuckerberg, Elon Musk and Jensen Huang.

Is the accord legally binding?

No, the accord is voluntary and does not carry legal penalties, though it states that some of its commitments could be codified into law or regulation in the future.

Why are critics sceptical of the agreement?

Critics point out that the participating companies wrote the accord’s principles themselves and will select their own external auditors, meaning there is limited independent oversight built into the process.

How does this accord relate to global AI regulation efforts?

It comes shortly after twenty countries and the European Union called for coordinated international AI oversight, an approach Trump has explicitly rejected in favour of US-led, industry-driven self-regulation.