Anthropic CEO Dario Amodei recently pushed back against claims that his overly pessimistic vision of AI and its future impact has fueled a negative public backlash. Amodei’s remarks were part of a response to investor Gavin Baker. Baker noted on both a podcast and on X that Amodei’s warnings about AI dangers have, to some extent, contributed to the anti-AI sentiment in the United States, particularly concerning data center construction. Baker wrote that, given Amodei has “lost the argument” in AI regulation discussions (Anthropic had advocated for some regulatory proposals, including a California bill requiring large AI companies to follow transparency rules), and considering “he is about to become CEO of one of the world’s most important companies,” “out of respect, I think he should try to be a more positive advocate for his industry.”
Baker is not alone in believing that public skepticism and even government crackdowns on AI are a natural reaction to dire warnings from some AI executives. However, in a series of posts, Amodei disagreed with the idea that his “messaging is too negative.” Instead, he said his writings strike a “fairly balanced consideration between risks and benefits,” and that he wrote his article “Machines of Loving Grace” precisely because he “felt the AI industry at the time wasn’t painting a sufficiently inspiring picture of how this technology will fundamentally make the world a better place.”
Despite this, Amodei acknowledged that “the public has a negative view of AI” and agreed “this is a major problem.” Where he disagreed was the notion that this negative sentiment “is primarily caused by him or any other AI leader warning about AI risks.”
Amodei stated: “I think fundamentally, this is a crisis of trust. Ordinary people lack trust; they don’t trust corporations, governments, or the tech industry, and they feel we are cooking up some new trick to harm them.” In fact, “trust” is a word that frequently appears in headlines about OpenAI CEO Sam Altman, and it is clearly a major challenge for other AI company CEOs like Amodei as well.
However, in Amodei’s view, this crisis has been brewing for decades, and the AI backlash is “merely its latest manifestation.” Amodei said: “I think the most accurate criticism of AI companies, including Anthropic, so far is that we have not delivered on the grand promises of benefiting the world. This is entirely our own problem, and I believe this is the criticism people should be making, rather than fussing over messaging and marketing tactics.” In other words, he noted that promising AI will cure cancer has “become more of a cliché than an inspiring vision”; what will truly change people’s perception of AI is “actually curing cancer.”
Turning to regulation, Amodei argued that Baker presented a “false choice” between “widely promoting AI without regulation” and “concentrating technology in the hands of a few companies through regulation.” Amodei said: “I know there is a simplified logic in Silicon Valley that ‘regulation = regulatory capture = power concentration,’ but I have always felt this is an oversimplified picture of the real world. To many outside this circle, regulation is seen as a constraint on corporate power and beneficial to ordinary people.”
Amodei added that he himself “doesn’t necessarily fully agree with this perspective,” but he said this is “precisely why Anthropic has always been extremely careful when formulating policy proposals.” He stated: “We have worked very hard to propose measures that would limit (slow down) the development of frontier AI companies while being beneficial to smaller competitors.”
Why propose these measures? Amodei said that in his view, “AI is structurally a technology that tends to concentrate power,” and while “open-source weights do help in this regard, they are far from a sufficient solution because they merely shift the concentration to those who have the most computing power and chips.” He continued: “In contrast, I believe that reasonable ‘industry rules of the game’ can simultaneously achieve three things: (a) address AI risks related to cybersecurity, biosecurity, and alignment; (b) institutionally constrain the power of frontier AI companies; and (c) preserve space for open-weight models while addressing their specific risks.”