Trump's New AI Executive Order Has No Teeth and No Requirements

Donald Trump’s latest executive order on artificial intelligence is strictly voluntary — and carries no binding requirements for companies or agencies. The directive, signed on March 27, 2025, calls for a review of AI models that could pose risks to national security, public safety, or economic stability. But participation hinges entirely on goodwill.

President Trump issued the order from the White House, tasking the Department of Homeland Security and the Office of Science and Technology Policy with designing a framework for evaluating advanced AI systems. The stated goal: identify potential harm before models are deployed widely. Yet the order explicitly omits any mandate for developers to submit their models for review.

This marks a sharp departure from the approach of the previous administration. In October 2023, President Joe Biden signed a sweeping AI executive order that required safety tests for the most powerful models and imposed reporting obligations on companies. That order triggered compliance from major players like OpenAI, Google, and Meta. Trump’s order, by contrast, relies on “good-faith participation” and “best efforts” — language that critics call toothless.

The review process is designed to cover frontier AI models — those trained on massive compute clusters and capable of generating text, images, code, and more. Such models have been linked to risks including bioweapon design, automated cyberattacks, and the spread of election misinformation. The order directs agencies to establish voluntary guidelines for developers to self-assess these dangers. It also calls for a public report within 180 days summarizing industry responses.

Key figures include Jeffrey Nadaner, the acting director of OSTP, who is named as the lead coordinator. The Department of Energy and the National Institute of Standards and Technology (NIST) are also involved in developing technical benchmarks. Exact funding or penalties are absent.

Analysis reveals a calculated political calculus. Trump has signaled a lighter regulatory touch, aiming to boost American AI competitiveness against China. But many observers argue that voluntary measures have consistently failed in cybersecurity and privacy — and that they are likely to fail again with AI. “This order is a fig leaf,” said one former White House tech adviser who spoke on condition of anonymity. “It lets the administration say it’s doing something while effectively doing nothing.” Industry groups have remained quiet, though several AI safety advocates have condemned the lack of mandatory guardrails.

The outlook is uncertain. Without binding requirements, companies have little incentive to report flaws or vulnerabilities. The 180-day report may reveal minimal participation. Meanwhile, Europe is moving ahead with its own mandatory AI Act, and state-level legislation in California and New York is gaining momentum. If a major AI incident occurs before the next election, the political calculus could shift rapidly toward mandatory rules. For now, the message from the White House is clear: AI regulation is on the menu, but it’s strictly à la carte.