Anthropic CEO Dario Amodei is pushing back on the claim that his warnings about artificial intelligence have helped turn the public against the technology. His argument is narrower and more uncomfortable for the industry: the backlash is not mainly a messaging problem, but a trust problem.
The exchange began after investor Gavin Baker criticized Amodei’s public posture on AI, including on the All-In podcast and on X. Baker argued that warnings from AI executives have contributed to opposition in the United States, especially around data centers, and said Amodei should become a more positive advocate for the industry he helps lead.
The dispute over AI messaging
Baker’s criticism centered on the idea that AI leaders have made their own political environment harder. He argued that Amodei has “lost the argument” on AI regulation, while noting that Anthropic has backed some regulation, including a California bill that imposes transparency requirements on large AI companies.
Amodei rejected the view that his public comments have been unusually negative. In a series of posts, he said his work has been “about equally balanced between risks and benefits.” He also pointed to his essay “Machines of Loving Grace,” saying he wrote it because he “didn’t feel the AI industry was painting an inspiring enoug h picture of how the tech nology could radically transform the world for the better.”
That response matters because it separates two arguments that are often treated as one. One question is whether AI executives should speak openly about cyber, bio, alignment, and concentration risks. Another is whether the public’s suspicion of AI comes from those warnings, or from deeper doubts about the institutions building and governing the technology.
Amodei says the backlash is about trust
Amodei acknowledged that “the public has a negative view of AI” and said he agrees “this is a big problem.” But he disagreed that the negative view is “primarily caused” by him or by other AI leaders talking about risks.
“I think it is fundamentally a crisis of trust,” Amodei said. “I think that ordinary people don’t trust companies, governments, or the tech industry and always suspect that we are cooking up some new way to screw them over.”
In his framing, the AI backlash is not an isolated reaction to one company, one chief executive, or one set of interviews. He described it as part of a trust problem “decades in the making,” with AI becoming “just the l atest iteration of it.”
That distinction is important for the industry. If the issue is only messaging, then the answer is better marketing, more optimistic language, and fewer warnings. If the issue is trust, then the answer has to involve proof: visible benefits, credible limits, and a clearer sense that powerful companies are not simply asking the public to accept another major technological shift on faith.
Promises are not enough
Amodei said the strongest criticism of Anthropic and other AI companies is not that they have talked too much about danger. It is that they have not yet shown enough concrete value to match their biggest claims.
“I think by far the most accurate criticism of AI companies including Anthropic is that we haven’t yet delivered on our big promises to benefit the world,” Amodei said. “That is totally on us, and I think it’s the criticism you should be making, instead of all this stuff about messaging and marketing.”
He used cancer as an example of the gap between inspiring claims and actual outcomes. Saying AI will cure cancer, he argued, has become “more a cliche than it is inspiring”; what would change views of AI would be “actually curing cancer.”
That point cuts against both simple optimism and simple pessimism. Amodei is not saying the industry should stop describing what AI might do. He is saying public confidence will not be rebuilt by repeating the largest possible promises. It will depend on whether AI companies can turn those promises into outcomes people recognize as real benefits.
The regulation debate
The argument also touched regulation. Amodei said Baker was presenting “a false choice” between broad AI distribution with no regulation and a regulated system that leaves the technology concentrated among a few companies.
Amodei criticized what he called a common Silicon Valley shorthand: “regulation = regulatory capture = concentration of power.” He said that view is too simple, while noting that many people outside that bubble see regulation as a way to limit corporate power and help ordinary people.
He did not fully endorse that outside view either. Instead, he said Anthropic tries to make policy proposals carefully, with a specific goal: “We try very hard to make proposals that disadvantage (slow down) frontier AI companies while *advantaging* smaller competitors.”
That is central to his case for regulation. Amodei said AI is “structurally” a technology that tends to concentrate power. He also argued that open-weights models help only partly, because they can shift concentration toward those with the most compute and chip.
In his view, the right rules could do several things at once:
- address AI’s cyber, bio, and alignment risks;
- institutionally constrain the power of frontier AI companies;
- leave room for open-weights models while dealing with the risks they create.
What the AI industry still has to prove
The exchange shows how much of the AI debate now turns on credibility. Baker’s view is that leaders like Amodei should make the case for AI more positively, especially as backlash builds in the United States and around data centers. Amodei’s answer is that public opinion will not be fixed by sounding more upbeat.
Instead, he is pointing to a harder standard. AI companies have to show that their technology benefits the world, that their policy proposals do not simply protect incumbents, and that their warnings about risk are not just another form of corporate positioning.
That leaves the industry with a practical problem. It cannot rely only on promises, and it cannot assume that people will trust companies, governments, or the tech industry to manage the consequences. For Amodei, the path forward is not to stop talking about risk. It is to earn trust by delivering benefits and supporting rules that match the scale of the technology.