As Anthropic and the U.S. government battle over the safety of a new AI model, Verasight sought to understand where the American public stands on the increasingly salient issue of AI safety and potential regulation. In a polarized world, the new Verasight survey of more than 1,600 Americans, shows that Americans overwhelmingly support AI regulation, regardless of political identification.

Americans trust the Government to judge an AI model’s safety twice as much as they trust Anthropic

The strong bipartisan support for government-led regulation is reflective of Americans’ trust in the Government to judge the safety of AI models, and broad skepticism of the ability of AI companies to self-regulate. 

Over the past few months, AI companies, such as Anthropic, have recommended regulations that they claim are designed to protect Americans. Verasight has found that Americans distrust the motives behind AI companies' attempts to self-regulate. Only 9% of respondents believed that AI companies’ proposals to regulate are genuine attempts to protect the public from AI, while 43% of Americans believe they are primarily designed to benefit the companies. 

In contrast, when asked who should have final authority over AI safety decisions when there is a dispute, more than three times the number of Americans believe that the US government should have final say than believe AI companies should (49% of respondents to 15%). The result was split evenly across the political spectrum with 53% of Democrats and 52% of Republicans believing the US government should have final say when in disputes with AI companies. 

In light of their current public dispute, we asked Americans specifically who they trusted more to judge the safety of AI models, the Government or Anthropic. Twice as many Americans said they trusted the US Government to judge safety as did trust Anthropic (30% to 14%). The highly politicised nature of the dispute and the Trump administration’s outspoken position on the topic, did not seem to deter Democrats from saying they trusted the government on issues of AI safety more than Anthropic, However, Republicans were the group most likely to trust the government rather than Anthropic (38% trusting the government and only 9% trusting Anthropic).

Almost 90% of Americans support the regulation of AI technology

When asked by Verasight about proposed AI regulations, 89% of Americans agreed that AI companies should be required to publicly disclose the results of safety testing for their most powerful AI systems. Similarly, 89% agreed that AI companies should be required to have their most powerful models reviewed by independent third parties. Most notably, Democrats and Republicans support these regulations almost equally - 90% of Democrats and 88% Republicans support public disclosure of safety testing for AI models, and 91% of Democrats and 87% of Republicans support independent review of these models. The level of bipartisan support, rare in today’s political climate, is telling of the unique position that AI technology occupies in the American political landscape. 

Americans from both parties are also meaningfully unified in the belief that the Government should have the authority to block the release of an AI system if it is found to pose a risk. We found that 81% of respondents supported this, with 83% support from Democrats and 80% from Republicans. 

69% of Americans support AI stock transfer proposal

Verasight also asked Americans whether they supported proposed legislation that would require artificial intelligence companies to transfer 50% of their stock to a fund owned by the American public. This proposal was originally suggested by Senator Bernie Sanders. Verasight randomly assigned respondents to see one of two questions, the first explicitly tying the proposed legislation to Senator Sanders, and the second not naming him. When he was not named, 69% of Americans supported AI stock transfer. While more Democrats supported it (82%), the proposal still gained significant traction among independent and Republican voters (77% and 59%, respectively). 

When Senator Sanders' name was explicitly linked to the proposal,  64% of respondents supported the proposal. However, even when tied to Sanders, 48% of Republicans still supported it. This once again demonstrates the unique potential for bipartisan support. The proposal was supported largely evenly across education and income levels. It was most significantly supported by those in the 18-29 age group, with 81% support. 

With the valuations of AI companies rapidly growing, bipartisan support for regulation and stock transfers raises the question of if these risks are priced in appropriately by investors. The full report is available here and you can contact us to learn more about how Verasight helps academic, media, and F500 researchers understand American public opinion. 

Source: Verasight Survey of 1,690 U.S. adults conducted June 18 - 29, 2026. The margin of error, which accounts for the design effect and is calculated using the classical random sampling formula, is +/- 2.8%.