The Trump administration arrived in office vowing to liberate the AI industry from regulatory constraints. Instead, it has constructed an invisible policy framework that dictates how companies operate, despite insisting it opposes formal regulation.
The contradiction runs deep. The White House rescinded Biden-era AI requirements and publicly championed a deregulatory stance. Yet behind closed doors and through executive action, it has become the de facto rule-maker for an industry worth hundreds of billions of dollars.
What has emerged looks nothing like traditional regulation. There are no published standards, no formal rulemaking process, no clear guidance posted for companies to follow. Instead, the administration operates through case-by-case deals, voluntary frameworks, and selective interventions that leave industry executives guessing what comes next.
Export controls on advanced AI models have become a primary tool. So have voluntary testing frameworks and government procurement guidelines. Anthropic, OpenAI, and other leading labs are currently negotiating compliance with a new executive order establishing a framework for government review of certain AI systems. Anthropic separately is haggling with the administration over whether export restrictions on its latest models can be lifted. These negotiations happen outside public view, shaped by personalities and broader politics as much as by written policy.
The General Services Administration is drafting additional rules around how large language models can handle government data, setting privacy and security standards that companies must meet to win federal contracts. Again, these are not laws Congress passed. They are bureaucratic moves executed by a single agency.
Congress was supposed to be the lawmaker here. A bipartisan AI safety bill circulated in the House, but Capitol Hill has stalled as midterm politics loom. That vacuum has given the Trump administration an opening. Lacking clear national AI rules from Congress, it has focused on overriding state AI laws, managing national security concerns around advanced models, controlling how AI enters government procurement, and monitoring the economic power of the biggest AI companies.
The uncertainty this creates is profound. Companies cannot simply read a statute and comply. Instead, they navigate an opaque system where rules shift with each executive order or negotiation. Industry executives spoke with Trump at this week's G7 summit about establishing a global forum for AI standards, a sign they are desperate for clarity.
What happens in Washington matters globally. The U.S. houses the world's most advanced AI models. When the Trump administration makes decisions about export controls, testing requirements, or government use, those decisions ripple across borders. Foreign leaders at the G7 spoke of building tech sovereignty to reduce dependence on American companies, yet they know they cannot ignore the leading models developed here.
The administration may resist calling this regulation. It certainly is not Europe's AI Act or the UK's Online Safety Act. But through procurement rules, export controls, voluntary testing, and targeted interventions, the Trump administration is setting the terms on which AI companies operate in the world's largest market. That power to shape an industry through the shadows is, for better or worse, regulation all the same.
Author James Rodriguez: "Without clear rules in writing, companies are navigating politics instead of policy, and that chaos spreads worldwide."
Comments