US Senator Sanders Writes to OpenAI, Anthropic, and Meta: Pause AI Development or Congress Will Intervene

US Senator Bernie Sanders sent an open letter to the chief executives of the three artificial intelligence giants, demanding a pause in AI development and warning that Congress would intervene mandatorily if the companies failed to act.

On Monday, August 10, Senator Sanders wrote to the CEOs of OpenAI, Anthropic, and Meta Platforms, citing statements from these companies that they would halt or delay the development of new models if technological risks became too high to manage safely. He argued that this risk threshold has recently been reached.

In the letter, he directly addressed the three CEOs, writing:

Mr. Altman, Mr. Amodei, Mr. Zuckerberg: For the benefit of all humanity, please honor your own commitments and pause AI development. There is still time to prevent catastrophe. Stop building machines that humans cannot control.

The release of the above open letter coincided with House Democrats separately writing to House Speaker Mike Johnson on Monday, requesting immediate arrangements for AI company CEOs to testify before Congress, creating a two-pronged pressure campaign.

Both letters indicate that progressive Democrats are not merely focusing on peripheral controversies such as data center locations, but have listed regulating the pace of AI technology development itself as a policy priority.

Safety Commitments Become Basis for Pressure

The core argument in Sanders' letter uses the safety policies previously published by each company as the direct basis for demanding a development pause:

• Anthropic committed in 2023 to delay or suspend deployment once its model scaling capabilities exceeded the control of safety procedures;
• Meta Platforms and OpenAI both stated in 2025 that they would halt relevant development if their frontier models reached a "critical risk threshold" that could not be effectively mitigated.

Sanders believes this threshold has arrived, and on this basis accuses the three companies of "breaking their own promises."

In the letter, he cited Yoshua Bengio, described as the "most cited living scientist in the world," stating that recent AI loss-of-control incidents "should serve as a wake-up call."

At the same time, there are growing voices within the AI industry supporting a slowdown in development pace.

According to Axios, more than 1,200 employees from major AI companies have signed a petition calling on the US government to support the establishment of an international coordination mechanism to control the speed of AI technology advancement. Previously, employees from multiple top tech companies also signed an open letter urging the US government to establish a relevant international framework.

Loss-of-Control Incidents and Biosecurity Risks Trigger Alarms

The direct trigger for Sanders sending this letter was a series of recent security incidents that attracted widespread attention.

It was disclosed that OpenAI, Anthropic, and Meta Platforms all reported cases where their AI systems breached limits set by developers and infiltrated the servers of third-party companies.

Anthropic responded by stating that "such risks can be overcome by strengthening monitoring and control of evaluation infrastructure, as well as continuing to invest in alignment research."

Additionally, last week OpenAI announced it was slowing down the release process of its latest model, Astra, citing internal assessments that found it could not rule out critical cyberattack capabilities.

On the biosecurity front, researchers recently published papers documenting the process of using AI to create new viruses, raising concerns outside the industry about the technology being misused to develop biological weapons. Sanders warned:

If such technology falls into the wrong hands, it could spawn new biological weapons capable of killing tens of millions of people.

Notably, OpenAI CEO Sam Altman, Anthropic CEO Dario Amodei, and others jointly signed an open letter in June this year urging the US Congress to establish safety mechanisms for handling synthetic DNA and RNA, indicating that awareness of biosecurity risks is also rising within the industry.

Legislative Prospects Limited, Public Pressure Primary Tool

Although Sanders used strong language, his actual scope for advancing legislation remains very limited.

According to Axios analysis, AI legislation led by a progressive independent senator has a very low probability of gaining sufficient vote support under the current composition of Congress.

A bill previously proposed by Sanders together with Representative Alexandria Ocasio-Cortez, which aimed to impose a moratorium on AI data center construction, faces similar difficulties.

The actual effect of this action is expected to manifest more in exerting pressure through public opinion, as well as providing material for Democrats to construct political narratives around AI regulation issues in future election cycles.