Why AI Firms Face a New Risk in Trump's Voluntary Vetting Order

The Executive Order That Banned Licensing but Opened a Different Door

President Donald Trump's June 2 executive order on artificial intelligence directs the creation of a voluntary program to vet "covered frontier models" for cybersecurity purposes. The document explicitly prohibits any licensing scheme for AI models. That provision seemed unlikely until last month, when National Economic Council director Kevin Hassett floated the idea of an FDA-style approval process for AI systems. Hassett said the administration was "studying possibly an executive order to determine how future AIs that also potentially create vulnerabilities should go through a process so that they're released in the wild after they've been proven safe, just like a FDA-approved drug."

The FDA reference was a red flag for an industry that views the agency as the epitome of safety-ism and bureaucratic excess. Trump himself has called the FDA's operations "slow and burdensome." By explicitly barring a licensing regime, the order removed the single most feared regulatory outcome. The document's omissions create a different set of risks for AI developers.

What the Order Left Unspecified

The executive order does not define which models qualify as "covered frontier models." It does not specify the criteria under which models will be evaluated. It does not explain how the administration will "select trusted partners" for the third-party vetting process. Some vagueness is inevitable in any document that directs agencies to build a new program. The breadth of the discretion delegated to bureaucrats is the core risk for the industry.

Mutable standards, changed at the whim of regulators, would rob AI firms of certainty and subject them to the will of partisan or self-interested administrators. The safety-ist faction within the administration may have yielded on the licensing question in the drafting of the executive order. The national security bureaucracy, not distinguished for its attention to civil liberties, will have ample chances for mischief as the order is implemented.

The Voluntary Problem: When Compliance Becomes Compulsory

The order describes the vetting process as "voluntary." Any student of political economy knows that such standards often prove compulsory in practice. Guidance documents and bureaucratic suggestions carry within them a veiled command. The experience of the financial-services industry demonstrates this pattern. James Madison noted that "an advisory government is a contradiction in terms."

The increasing entanglement of leading AI firms with the federal government through government contracts creates a colossal incentive structure guiding innovators to comply for survival's sake. Opting out may not be a reasonable option for AI firms. "Comply, or else" is not infrequently the information conveyed between the lines of "voluntary" regulatory provisions. This administration has already revealed its taste for retribution against non-compliant firms in its pursuit of Anthropic, beyond any legitimate state interest.

The Retribution Precedent

Anthropic's treatment signals a pattern. The administration has demonstrated willingness to act against firms that do not align with its preferences. For AI developers weighing whether to participate in the voluntary program, the Anthropic case provides a concrete data point. The cost of non-compliance may extend beyond lost government contracts to include active regulatory friction.

The Regulatory Molding Problem