Letters made public on Tuesday, October 6, Eastern Time, show that U.S. Democratic Senators Elizabeth Warren and Richard Blumenthal sent a joint letter this week to Trump administration officials including Treasury Secretary Bessent and Commerce Secretary Lutnick, demanding an explanation of whether tech giants have influenced the government's AI regulatory policy through lobbying, and requesting disclosure of the specific contents 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" rather than addressing the security and national security risks posed by increasingly powerful AI systems. They demanded that the government "stop putting industry profits above the interests of the American people."

Republican Senator Bernie Moreno, meanwhile, sent a letter this week to Anthropic CEO Dario Amodei, warning against taking an "alarmist" attitude toward AI. Moreno argued that there is a contradiction in Anthropic preparing for an IPO and hoping the American public will share in the company's growth dividends while simultaneously warning that its technology could ultimately lead to human extinction. He urged Amodei to place greater emphasis on the potential of superintelligence to improve the lives of Americans, strengthen national security, and address AI threats.

From Mandatory Regulation to Voluntary Framework: Democrats Question How "Big Tech Lobbying" Changed Policy

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

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

But the two senators cited media reports in pointing out that after executives from companies including Google, Anthropic, and OpenAI voiced opposition to the draft, the government changed the mandatory framework to a voluntary mechanism and narrowed its scope. Subsequently, after Meta CEO Zuckerberg and xAI CEO Musk reportedly expressed concerns to Trump, the testing window was further shortened from 90 days to 30 days, and was ultimately incorporated into Executive Order 14409 signed in June.

The senators further accused the recently promoted "White House Accord on Super Intelligence" of being a self-regulatory arrangement led by the tech industry. The letter cited reports stating that Zuckerberg played a "central role" in the formation of the accord, while Nvidia CEO Jensen Huang helped secure support from other companies. Trump described the accord as having "moral force," but Warren and Blumenthal argued that it still cannot replace genuinely binding government regulation.

Demand for Disclosure of Records of Communication Between Companies and Government

The two senators demanded that the government answer a series of questions by October 19, including disclosing in-person meetings and communications between tech companies and government officials regarding Executive Order 14409, as well as different versions of the draft, and providing a set of 31 questionnaires reportedly sent by ONCD to the industry along with the companies' responses.

They also pressed on whether independent evaluators, consumer organizations, critical infrastructure operators, and academic experts outside the industry were consulted when formulating the regulatory framework, and demanded that the government explain exactly what capabilities and risks are assessed in pre-deployment testing of frontier AI models, what benchmarks are used, and what measures the government would take if a model fails testing.

This marks the latest in a series of recent pressures by the two lawmakers on the Trump administration's AI regulatory policy. On September 28, they demanded that the Treasury Department explain the government's secretive, voluntary testing mechanism for advanced AI models, and questioned why the government had not yet announced a federal investigation into the relevant companies following an AI agent network attack incident related to OpenAI.

Republican Senator Criticizes Anthropic for "Alarmism"

It is worth noting that the political debate surrounding AI safety is not a simple partisan divide.

According to Axios, Republican Senator Moreno sent a letter on October 5 to Anthropic CEO Amodei, criticizing his "alarmist" attitude toward AI risks and warning against making fear-inducing statements.

Moreno specifically seized on the timing of Anthropic's preparations for an IPO, questioning that the company on one hand wants ordinary American investors to share in its growth dividends while on the other hand warning that AI could ultimately lead to human extinction. He asked Amodei to place greater emphasis on the potential of superintelligence to improve livelihoods, strengthen U.S. national security, and address AI attacks. Moreno also said that addressing frontier AI risks does not mean pausing development or abandoning America's technological leadership.

From this perspective, the current debate in Washington over AI policy is revealing an increasingly clear main line: Democrats are more concerned about tech giants influencing policy, insufficient regulation, and AI safety risks, while the Trump camp and some Republicans place greater emphasis on avoiding "AI doomsday" narratives that hinder innovation and U.S. competitiveness. Anthropic and Amodei are gradually becoming an important focal point in this policy divergence.