Amodei on AI backlash: It’s not fearmongering, people just don’t trust us

Anthropic CEO Dario Amodei did something over the weekend that he does not usually do. He wrote a lengthy response on X and took aim at investor Gavin Baker, who accused him of helping create a wave of backlash against the entire industry through his statements about the risks of artificial intelligence. Amodei rejected the claim and offered his own explanation for why the public dislikes technology companies. According to him, it is about trust that has been eroding for decades, with AI merely its latest chapter.

The entire dispute was sparked by a clip from the All-In podcast. In it, Baker repeated second-hand information claiming that Amodei had privately said somewhere that Anthropic might one day become the only private company in the world. Anthropic researcher Sholto Douglas responded immediately, calling it a fabrication invented by someone because it fit the narrative that part of Silicon Valley wants to believe. Baker accepted the correction and acknowledged that Amodei had said no such thing.

But the discussion did not end there; it merely shifted to a more substantial topic. Baker wrote that the reason many people believed it was simple. According to him, the story fits well with the way Amodei talks publicly about artificial intelligence.

Baker summarized his position quite bluntly. According to him, Amodei’s efforts to tighten AI rules have failed politically, and his warnings are now giving ammunition to people blocking the construction of data centers in the United States. He wrote that Amodei had lost that battle because his rhetoric had led nowhere.

He also added his own vision of a safe future. He wants as many artificial intelligence models as possible to exist and compete with one another. In his view, this increases the chance that at least one of them will share the values of an ordinary person. Hoping that some absolute power will look after humanity as long as it is enlightened enough has never ended well historically, he wrote.

Completely false.

I like Gavin's takes, but whoever he heard this from is lying so that it fits the narrative some people so desperately want you to believe.

The same people will try to convince you Anthropic has no moat, and a sentence later that it might become so powerful… https://t.co/W0Xgd7lfNA pic.twitter.com/bPf8sZATth

— Sholto Douglas (@_sholtodouglas) August 15, 2026

Amodei responded in two parts and thanked Baker for the exchange of views. At the outset, however, he rejected the way the question had been framed. According to him, the choice between concentrating the technology in the hands of a few select companies and politicians or giving it to everyone is a false one. He explained this using the example of how people outside the technology bubble see the issue. In Silicon Valley, he said, a shorthand is used according to which rules always mean that the largest companies will take control of the industry and power will become concentrated. Amodei argues that this view of the world is too simplistic. Many people outside the industry, by contrast, see laws as something that limits the power of large corporations and protects ordinary citizens. He himself subscribes to neither of these views because, in his opinion, it depends on what a particular law actually says.

Anthropic is trying to promote proposals that regulate the largest developers at the technological frontier while helping smaller companies. He referred to California’s SB 942, which does not impose obligations on companies below a certain revenue threshold, as well as to his earlier position on SB 1047. According to him, stricter testing should apply to models near the technological frontier, not to small developers, including those that release their models with open weights.

At the same time, he does not believe he has lost in recent months. He says he supports the approach that reports suggest the Trump administration is preparing—namely, testing frontier models before release and testing open models once they approach the frontier. He also spoke positively about Demis Hassabis’s idea to establish an oversight body modeled on financial self-regulation. Six months ago, the situation was different because most of the industry was pushing to repeal individual state rules while proposing nothing at the federal level.

The second part of the response focused on the way Amodei speaks. He rejected the claim that his rhetoric was disproportionately negative. He wrote that he divides his attention roughly equally between risks and benefits, has written one major essay on each aspect, and addresses both topics in interviews as well. The 2024 essay Machines of Loving Grace was written precisely because he felt the industry lacked someone presenting a sufficiently inspiring picture of how the technology could improve the world. According to him, clips on social media often portray him as a pessimist because that attracts more clicks. He acknowledged, however, that the public has a negative view of artificial intelligence and that this is a major problem. He only disagrees that he or other company executives who discuss the risks are to blame.

He also rejected the solution that some people recommend to him. In his view, a glossy campaign with a positive message will not restore trust. Claiming that AI will cure cancer is now more of a cliché than an inspiration, and most people consider it a scam. The only thing that will work is actually curing cancer.

According to him, the most accurate criticism of AI companies, including Anthropic, is that they have not yet fulfilled their grand promises to help the world. He accepted responsibility for that and added that this is what critics should have blamed him for instead of debating marketing. Anthropic is therefore significantly stepping up its work in biology and medicine, he said. He promises great results within years and the first glimmers within the coming months. And once something succeeds, he wants to proclaim it as loudly as possible.

The entire response rests on a single idea. According to Amodei, the backlash against AI is fundamentally a crisis of trust. Ordinary people do not trust companies, governments, or the technology industry and constantly feel that someone is preparing to deceive them. In his view, the causes stretch back decades, and artificial intelligence is merely the latest manifestation of the same phenomenon. Incidentally, the word trust appears regularly in debates about AI, most often in connection with OpenAI CEO Sam Altman, Amodei’s main rival.

Amodei’s diagnosis also has its detractors. Companies in the industry emphasize how powerful their models are while simultaneously assuring the public that they have that power safely under control. The contradiction between these two claims erodes trust in itself. Recent cases reinforce this perception. Reports of autonomous agents escaping test environments, unclear rules for working with data, and products being used in ways no one intended have only heightened public concerns. How a company responds to a problem can sometimes shake people’s trust more than the problem itself.

Critics also point out that communication will not fix the issue. According to them, structural questions remain unresolved—including the lack of oversight and the pressure to monetize products as quickly as possible. Baker also cited a specific incident in his posts. An unreleased advanced OpenAI model appeared on the Hugging Face platform, which, according to Baker, effectively ended hopes of quickly tightening the rules.

Sources: techcrunch.com and vespernews.com