Trump Killed AI Safety Rule After Taking Millions From Industry

NEWS & RESEARCH

In September 2024, the Biden administration proposed a rule that would have required artificial intelligence (AI) companies to file quarterly reports telling federal regulators what kinds of experiments they were running on their most advanced models, so the government could monitor threats to national security and public safety. The rule drew objections from AI companies including OpenAI and Anthropic, and it had not been finalized by the time Biden left office. Shortly after returning to the White House in 2025, President Trump signed an executive order revoking Biden's directive on AI disclosure. This decision came just nine days after Trump's super PAC, MAGA Inc., received a $5 million donation from a company chaired by Sam Altman, the CEO of OpenAI. However, even after Trump's order, Biden's original proposal remained on the Commerce Department's rulemaking docket for nearly a year with no action taken. During that time, Trump's political operation collected millions more from the AI industry. Altman met with Trump at least twice while the rule was stalled, and two of his company's investors—Marc Andreessen and Ben Horowitz—donated a combined $9 million to MAGA Inc. And then came the biggest AI-related donation of all, as OpenAI cofounder and President Greg Brockman and his wife gave $25 million to MAGA Inc. in September 2025. Three months later, the AI disclosure rule was officially withdrawn from the federal regulatory agenda.

SOURCES: The Lever | The New Republic

ANALYSIS & OPINION

With the disclosure rule gone, federal oversight of the most powerful AI models depends on companies volunteering for it. In June 2026, Trump signed another executive order on AI security that requires no disclosure at all; according to a Congressional Research Service summary, it asks companies to "participate in a voluntary review process that gives the government a window of 30 days to examine new systems before such models are released." However, there is little reason to expect the industry to opt in to the same type of scrutiny it fought so hard against. But perhaps even more troubling is that the death of the Biden-era rule looks like pay-to-play politics. When a safety rule an industry opposed disappears after that industry's leaders bankroll the president's super PAC, the public cannot know whether the decision was made on the merits or paid for. In any event, given the magnitude of the threat posed by AI, in the case of injury due to its use, the AI company should be strictly liable for the harm.

SOURCES: The Lever | The New Republic | Reuters | New York Times

HOW TO FIX IT

Federal action:

• Pass a law requiring AI companies to report to federal regulators, so that disclosure rules cannot be undone by executive order. Several existing proposals would do this, including the FRONTIER Act, which would require transparency reporting and independent audits; the Artificial Intelligence Risk Evaluation Act, which would require a federal evaluation before advanced systems are deployed; the AI Incident Reporting Act, which would require developers to report serious AI safety incidents to the federal government; and the AI Systems Transparency Act, which would require companies to disclose how their models work and what safeguards they have. These would restore AI oversight but would not address the pay-to-play problem.

• Congress should create an AI commission to impose licensing requirements on AI models as it does for other dangerous products, like cars, airplanes, and drugs.

• Pass the Abolish Super PACs Act, which would cap contributions to super PACs at $5,000. This would have cut the major donations made by AI industry players to a tiny fraction of their original amounts, sharply reducing the financial leverage that makes pay-to-play possible. However, Congress would also have to close dark money loopholes to stop wealthy donors from simply finding other ways to give. The DISCLOSE Act of 2026, which would require groups that spend more than $10,000 in an election cycle to reveal their major donors, would be a step in the right direction. 

• Pass a bill that would require agencies to publish a written justification whenever they withdraw a proposed rule, along with a public log of every meeting and contact with affected companies, including at the White House, and a list of large donations from those companies' leaders. This would not stop an administration from killing a rule for its donors, but it would put the money, the meetings, and the decision on the public record.

• Urge Congress to investigate the withdrawal of the AI disclosure rule, including why the Commerce Department dropped it and what contact the administration had with OpenAI and other AI industry donors.

State action:

• Pass laws modeled on California's SB 53 and New York's RAISE Act, which require frontier AI developers to publish safety plans and report serious incidents, so oversight does not depend on federal action alone.

• Pass state laws clarifying that AI companies are strictly liable for harm caused by their products without a plaintiff needing to show that the company either intended to cause harm or was negligent. As between a company that profits off the dissemination of AI and an innocent victim harmed by AI, the equities demand that the company make the injured party whole. This is how other inherently dangerous activities are treated by civil law, such operating a nuclear power plant or storing toxic chemicals.

Litigation:

• Support Protect Democracy Project v. Office of the National Cyber Director, a lawsuit filed in September 2026 seeking to force four federal agencies to disclose how the administration's voluntary AI review framework works, which companies participate, and how they are chosen. A win would not restore mandatory reporting, but it would reveal whether the framework favors certain companies.

Legislation: H.R. 9925 - FRONTIER Act | S. 2938 - Artificial Intelligence Risk Evaluation Act | H.R. 9477 - AI Incident Reporting Act | S. 5471 - AI Systems Transparency Act | H.R. 2352 - Abolish Super PACs Act | H.R. 7802 - DISCLOSE Act of 2026 | SB 53 - Transparency in Frontier Artificial Intelligence Act (California) | S6953B - RAISE Act (New York)