Republicans criticize AI danger claims, Democrats question the influence of "big tech" on the Trump administration's AI policy and demand disclosure of regulatory framework amendment procedures

A public letter released on Tuesday, October 6, Eastern Time, shows that U.S. Democratic Senators Elizabeth Warren and Richard Blumenthal sent a joint letter this week to Treasury Secretary Benson, Commerce Secretary Lutnick, and other senior officials in the Trump administration, requesting clarification on whether tech giants have influenced the government's AI regulatory policies through lobbying. They also asked for disclosure of the details of the relevant regulatory framework and testing mechanisms.

The two senators stated directly in the letter that the Trump administration is “catering to the interests of big tech CEOs” instead of addressing the growing safety and national security risks brought about by powerful AI systems. They demanded that the government “stop putting industry profits above the interests of the American people.”

Republican Senator Bernie Moreno, on the other hand, sent a joint letter this week to Anthropic CEO Dario Amodei, warning against taking an “alarmist” attitude toward AI. Moreno argued that while Anthropic prepares for its IPO and hopes that the American public will share in its growth dividends, it also warns that its technology could eventually lead to human extinction, which he sees as contradictory. He urged Amodei to place greater emphasis on the potential for superintelligence to improve the lives of Americans, enhance national security, and address potential AI threats.

From Mandatory Regulation to Voluntary Framework: Democrats Question Whether “Big Tech Lobbying” Has Changed Policy

The letter focuses on tracing the process by which the Trump administration adjusted the AI regulatory framework from May to June this year.

Warren and Blumenthal stated that the plan originally drafted by the Office of the National Cyber Director (ONCD) in May required a mandatory 60-day government evaluation of advanced AI models, followed by a 75-day limited user testing period. If a model was deemed to pose unacceptable risks, the government could block its deployment.

However, the two senators cited media reports that, after executives from Google, Anthropic, and OpenAI objected to the draft, the government shifted from a mandatory framework to a voluntary system with a narrower scope. Subsequently, after Meta CEO Mark Zuckerberg and xAI CEO Elon Musk reportedly expressed concerns to Trump, the testing window was further shortened from 90 days to 30 days, finally being incorporated into Executive Order No. 14409, signed in June.

The senators further accused that the recent White House Accord on Super Intelligence is also a self-regulatory arrangement led by the tech industry. The letter cited reports that Zuckerberg played a “central role” in forming the accord, while Nvidia CEO Jensen Huang helped garner support from other companies. While Trump claims the accord is “morally binding,” Warren and Blumenthal argue that it still cannot replace truly binding government regulation.

Request to Disclose Communications Between Industry and Government

The two senators asked the government to answer a series of questions by October 19, including disclosure of in-person meetings, communication records, and various drafts related to Executive Order No. 14409 between tech companies and government officials. They also requested a copy of the 31-question survey supposedly sent by ONCD to the industry, along with the companies’ responses.

They further inquired whether non-industry actors such as independent assessment organizations, consumer advocates, critical infrastructure operators, and academic experts were consulted when forming the regulatory framework. The senators also asked the government to clarify which capabilities and risks current advanced AI models are evaluated for prior to deployment, the benchmarks used, and what actions the government will take if a model fails such tests.

This marks a continued effort by the two senators to pressure the Trump administration on AI regulatory policy. On September 28, they had already asked the Treasury Department to explain the government’s secret, voluntary testing mechanism for advanced AI models, and questioned why, following an OpenAI-related AI agent cyberattack incident, there had not yet been an announcement of a federal investigation targeting the companies involved.

Republican Senator Criticizes Anthropic for “Alarmism”

It is worth noting that the political debate over AI safety is not simply a partisan divide.

According to Axios, Republican Senator Moreno sent a letter on October 5 to Anthropic CEO Dario Amodei, criticizing him for an “alarmist” stance on AI risks and warning against making statements that could incite panic.

Moreno especially focused on Anthropic’s upcoming IPO, questioning how the company could simultaneously seek for ordinary American investors to share in its growth dividends while also warning that AI could ultimately lead to human extinction. He called on Amodei to place greater emphasis on the potential of superintelligence to improve lives, strengthen U.S. national security, and address AI threats. Moreno also remarked that addressing advanced AI risks does not mean pausing development or giving up America’s technological leadership.

As such, it is clear that Washington’s current debate over AI policy is forming an increasingly distinct main line: Democrats are more concerned about the influence of tech giants on policy, insufficient regulation, and the risks posed by AI safety, while the Trump camp and some Republican figures place greater emphasis on avoiding “AI doomsday narratives” that could impede innovation and American competitiveness. Anthropic and Amodei are gradually becoming a key focus of these policy disagreements.