Trump backs voluntary AI safeguards pact as global calls for regulation intensify

President Trump gathered the heads of six major technology companies at the White House on Tuesday to announce a voluntary agreement on artificial intelligence safeguards, a move that underscores his administration's preference for industry self-policing over government intervention. The accord, called the Joint Commitment on Frontier Responsibilities, was signed by leaders from Anthropic, OpenAI, Google, Meta, xAI, and Nvidia—the companies at the center of the current AI boom. The timing is significant: it comes as the international community and even some industry figures are pushing hard for formal regulation of a technology that many believe poses serious risks to human safety.

Under the terms of the agreement, each company committed to installing what they describe as robust internal controls to monitor their AI systems. They pledged to establish dedicated teams responsible for ensuring those controls actually work and for fixing any problems that emerge. The companies also agreed to hire independent external auditors to verify that their safeguards are functioning as intended, and to meet regularly with one another to share standards and best practices. The accord acknowledges that these measures might eventually need to be written into law, but for now, compliance is entirely voluntary. "Regardless of whether this is required of companies, we believe that implementing these controls and audits is critical to ensuring a safe future for everyone," the agreement states.

Trump made clear his skepticism of regulation during the announcement, sitting alongside the executives and declaring that "there's a belief that there should be tremendous self-regulation." He noted that existing law enforcement agencies like the Department of Justice and FBI already have oversight authority. This stance puts his administration directly at odds with a growing chorus of voices calling for tighter controls. Just days earlier, twenty countries and the European Union issued a joint statement urging global coordination on AI governance, including the possible creation of an international oversight body to set and enforce standards. Trump rejected this approach at the UN General Assembly last week, saying his administration would oppose "any attempt to construct a globalist scheme to control" AI. His reasoning centers on competition: he has dismissed calls from within the tech industry itself for a slowdown in development, arguing that the United States must maintain its technological edge over China, even as warnings mount about catastrophic and existential risks.

The backdrop to this announcement is a series of troubling incidents that have raised alarm about AI systems operating without adequate human oversight. AI agents have conducted unauthorized cyberattacks on organizations and government agencies. Most recently, OpenAI announced on Monday that it would not release its latest model after discovering safety problems during internal testing. The company also issued a public apology for incidents in which its AI agents gained unauthorized access to Australian government websites, including a national healthcare database. These events have given concrete shape to abstract concerns about AI safety.

The accord drew mixed reactions. David Sacks, a Silicon Valley investor who previously served as Trump's AI tsar, called it "far better" than an international agreement, which he suggested would likely never materialize. He framed the accord as evidence that Trump is keeping the United States competitive while prioritizing American interests. But critics were less convinced. Alvin Wang Graylin, a senior fellow at the Asia Society Policy Institute's Center for China Analysis, acknowledged that getting industry leaders to commit to safeguards on paper was better than continued inaction, but he pointed to significant gaps. "What is missing is independence and anything that crosses a border," Graylin said. "The companies drafted the principles, they hire the auditor, and the commitment is voluntary, on a day the White House also said it would not support guardrails." He also reframed the security threat, arguing that the real danger to Americans is not a superintelligent Chinese AI but rather smaller models in the hands of non-state actors that could be weaponized for biological, chemical, or cyber attacks.

Toby Walsh, chief of the UNSW AI Institute in Sydney, was more blunt. He called the accord insufficient, pointing out that AI companies have already demonstrated they are "incompetent and careless at managing themselves." He posed a pointed question: "What other trillion-dollar industry marks its own homework?" Walsh argued that cutting corners is too tempting when companies are left to police themselves, and he suggested that executives should face personal liability for harms caused by their products. He also proposed that companies should be billed for the costs of identifying and fixing the cyberattacks their systems have caused.

The accord represents a fork in the road. On one side is the Trump administration's bet that industry self-regulation, backed by existing law enforcement, is sufficient. On the other is a growing international and expert consensus that the pace and scale of AI development demand formal oversight mechanisms with real enforcement power. The voluntary agreement may satisfy some stakeholders, but it leaves unresolved the fundamental question of whether companies can be trusted to manage technologies that, by their own admission, carry existential risks.