US signs voluntary AI pact as EU pushes binding rules

Washington is betting on industry self-policing.

On 29 September, the White House hosted a lunch meeting with leaders from Meta, Google, Nvidia, and Anthropic. The result was not new legislation. It was a voluntary pact.

President Donald Trump signed an agreement described by officials as "morally binding" but legally unenforceable. The document pledges companies to implement solid safeguards. It also includes an executive order renaming the technology from "artificial intelligence" to "super intelligence" for federal purposes. The New York Times reported on the signing and the specific executive order details.

The strategy relies on the assumption that leading firms will police their own risks. Trump told reporters at the event that the industry would engage in "tremendous self-regulation." He rejected the idea of new federal laws, arguing that government rules would slow down American innovation. This approach marks a distinct departure from the regulatory path taken by other major economies.

The European Union disagrees.

On the same day, EU officials stated they would continue pushing for global AI safety rules. POLITICO.eu reported that Brussels intends to maintain its push for binding international standards. The divergence creates a fragmented global structure where the US relies on voluntary codes and the EU on statutory law.

The voluntary pact and its limits

CBS News reported that the agreement is "almost like a constitution" for the industry, but it lacks the teeth of federal law. Critics note that without legal enforcement, companies can choose how strictly to interpret the pledges. The term "morally binding" has drawn scrutiny from legal experts who question its practical effect on corporate compliance.

At the meeting, Trump rejected proposals for new safety laws. He argued that the technology is moving too fast for traditional regulation. Instead, he urged the public to trust the industry's self-policing. The Washington Post covered the president's defense of this light-touch strategy, highlighting his reliance on executive power rather than congressional legislation.

The executive order renaming the sector to "super intelligence" is a rhetorical move as much as a regulatory one. The Guardian noted that the document contained a spelling error, referring to the "Unites States." The rebranding aligns with the administration's broader narrative that the US is entering a new era of cognitive capability.

Industry executives attended the lunch to discuss regulation and competition with China. The New York Times analyzed the seating arrangement, suggesting it revealed who held the president's ear. The meeting included CEOs from major firms, but smaller players and civil society groups were absent. This exclusion has raised concerns about the representativeness of the voluntary framework.

Contrasting approaches in Brussels and Beijing

While Washington embraces flexibility, Beijing has moved faster on regulation. The Conversation reported that US tech leaders are urgently calling for rules, but China has already implemented them. Chinese regulations focus on application and content moderation, reflecting a different set of policy priorities. This creates a complex dynamic where US companies operating in China must comply with strict local laws, while US domestic policy remains voluntary.

The EU's position is firm. POLITICO.eu cited statements from EU officials confirming their intent to keep pushing for global safety rules. The bloc's AI Act is already in force, with phased implementation deadlines. This regulatory certainty appeals to some businesses that prefer clear guidelines over ambiguous voluntary pledges. However, it also imposes compliance costs that smaller firms may struggle to meet.

Analysts point out that the US approach carries risks. Without legal backstops, enforcement depends on corporate goodwill and public scrutiny. The Register reported on industry concerns about overlapping cybersecurity regulations, suggesting that even within the US, there is confusion about the regulatory environment. The voluntary pact does not resolve these ambiguities.

Economic implications and market reaction

The market reaction to the voluntary pact has been mixed.

Some investors welcome the lack of new regulations, seeing it as a boost for innovation. Others worry about the long-term risks of under-regulation, particularly in areas like data privacy and algorithmic bias. The volatility reflects the uncertainty surrounding the enforcement of the "morally binding" agreement.

The decision to rename the technology to "super intelligence" may have legal and contractual implications. Companies with existing contracts referencing "artificial intelligence" may need to update their language. This adds an administrative burden that was not part of the original regulatory debate. The focus on branding over substance has drawn criticism from legal scholars.

As the US moves forward with its voluntary framework, the rest of the world watches closely. The EU's commitment to binding rules and China's existing regulations create a patchwork of global standards. US companies operating internationally must navigate this complexity. The outcome of this regulatory divergence will shape global tech policy.

For now, the US has chosen trust over control. The voluntary pact is a test of whether industry self-regulation can manage the risks of advanced AI. The next few months will reveal whether the "morally binding" agreement holds weight in practice. The contrast with the EU's approach ensures that the debate will remain international in scope.