Amodei: AI Backlash Rooted in Broken Promises, Not Warnings

Investor Gavin Baker raised a direct challenge to Dario Amodei, chief executive of Anthropic, through the All-In podcast and social media platform X. Baker argued that Amodei’s persistent warnings about artificial intelligence risks have amplified public skepticism toward the technology in the United States and contributed to broader backlash, particularly opposition to data center expansion. Baker claimed Amodei has “lost the argument” on AI regulation and should become a more enthusiastic industry advocate rather than focusing on potential harms. He pointed to Anthropic’s support for California’s transparency bill, which requires large AI companies to disclose operational details, as evidence that even Amodei’s preferred regulatory approach has not gained traction in broader policy debates. Baker’s core contention was that someone in Amodei’s position, leading “one of the most important companies in the world,” should prioritize optimistic messaging about AI’s potential over cautionary warnings about its risks.

Amodei’s response directly challenged this premise. He argued that his public communications have been “about equally balanced between risks and benefits,” not skewed toward pessimism. To support this claim, he cited his essay “Machines of Loving Grace,” which he wrote not to warn about AI dangers but to fill what he perceived as a gap in industry narrative. The essay emerged, he explained, from his belief that the AI sector was failing to articulate a sufficiently compelling vision of how the technology could radically improve human life.

Trust Runs Deeper Than Messaging

Rather than accepting Baker’s diagnosis that messaging strategy drives public skepticism, Amodei reframed the entire problem. He acknowledged the reality that “the public has a negative view of AI” and described it as “a big problem.” But he rejected the notion that executives’ choice of tone or emphasis is the primary culprit.

“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.”

A Problem Decades in the Making

Amodei located the current AI backlash within a much longer historical arc. The skepticism people express about artificial intelligence is, in his formulation, “just the latest iteration” of declining institutional confidence that has been building for decades. This perspective fundamentally reframes the debate. If the underlying issue is structural distrust spanning numerous institutions over many years, then messaging strategy—no matter how carefully calibrated—will have limited capacity to shift public perception.

The mention of Sam Altman, CEO of rival company OpenAI, emerged naturally in this discussion of trust. Both executives have become symbolic figures in ongoing debates about whether AI companies can be trusted with such powerful technology. Yet Amodei’s analysis suggests the fundamental challenge is not specific to any individual leader or particular firm. Instead, it is systemic, affecting how people relate to technological and governmental institutions broadly.

The Industry’s Failure to Deliver

Amodei acknowledged one criticism he considers entirely valid. AI companies, including Anthropic, have not yet delivered on the transformative promises they have made to society. “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,” he 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.”

This concession reveals a crucial distinction in Amodei’s thinking. While he rejects the idea that his communications have tilted disproportionately negative, he fully accepts that the industry’s failure to produce benefits at meaningful scale provides the public with legitimate grounds for skepticism. The doubts people harbor are not primarily products of executives sounding alarms—they reflect the reality that those alarms, so far, have proven more credible than the promises.

The Regulation Debate

Read also: Google Makes Visible AI Watermarks Optional on Gemini · Anthropic explains how Claude’s watermarks will detect AI text · Tech Giants Face Natural Gas Price Risk Behind AI Expansion

Baker’s criticism extended beyond questions of messaging to the thorniest policy question facing the AI industry: regulation. He framed this as an inherent trade-off. Either AI develops with minimal government oversight, Baker argued, or regulatory frameworks inevitably concentrate power in the hands of a few dominant companies through what is commonly called regulatory capture.

Amodei rejected this as a false dichotomy. He criticized what he termed “Silicon Valley shorthand”—the assumption prevalent in technology circles that regulation automatically produces regulatory capture and concentrates corporate power. “I know that there’s a sort of Silicon Valley shorthand where regulation = regulatory capture = concentration of power, but I’ve always found this to be an overly simplified picture of the world,” Amodei said.

Different Views Outside the Tech Bubble

Amodei observed that people outside technology circles often understand regulation entirely differently. To many, regulatory frameworks function as tools to constrain corporate power and protect ordinary citizens, not as mechanisms for entrenching incumbent companies. While Amodei indicated he does not fully endorse that perspective either, he emphasized that the Silicon Valley assumption represents an incomplete picture of how regulation actually operates across different industries and historical contexts.

This observation provides important context for Anthropic’s actual regulatory positions. The company has supported specific measures, including California’s transparency bill requiring large AI companies to disclose operational details. Such advocacy would seem inconsistent with the view that all regulation inevitably harms smaller competitors while benefiting dominant firms.

Crafting Proposals That Don’t Entrench Power

When discussing how Anthropic develops regulatory proposals, Amodei described a deliberate strategy. The company attempts to craft policy ideas “that disadvantage (slow down) frontier AI companies while advantaging smaller competitors.” This articulation suggests Anthropic is trying to support regulation that would impose greater constraints on industry leaders while creating competitive openings for emerging firms.

Whether such regulation is practically achievable in the political economy of technology remains uncertain. But Amodei’s description indicates that Anthropic does not view regulation and competitive fairness as inherently opposed. The company’s stated goal is to support rules that advance public interests without allowing those rules to entrench Anthropic’s own market position.

What the Communication Record Actually Shows

Amodei provided specific evidence when defending his claim that his communications have not systematically emphasized risks over benefits. He had written one major essay articulating risks and another articulating benefits. By his account, the distribution of his major published pieces demonstrates balance. Even in interviews where he discusses potential harms from AI, the broader body of his work maintains equilibrium.

The “Machines of Loving Grace” essay exemplifies this in Amodei’s analysis. He wrote it not to minimize AI risks but because he felt the industry was neglecting to articulate a genuinely inspiring vision of what the technology could accomplish. The essay represented an attempt to fill a perceived gap in industry narrative rather than an effort to offset what Amodei considers responsible discussion of challenges and potential dangers.

The Fundamental Divide

The exchange between Baker and Amodei reflects a more basic disagreement about what drives AI skepticism. Baker’s position, shared by others in the industry, attributes much of the backlash to how executives communicate—the choice to emphasize risks, use cautionary language, and voice concerns about potential harms. In this view, more optimistic messaging would reduce public anxiety and facilitate smoother development and deployment of AI systems.

Amodei’s diagnosis points to different sources. Public skepticism, he argued, stems not primarily from messaging choices but from a real gap between promises and delivery. AI companies have made sweeping claims about the transformative potential of their technology. Rebuilding public confidence requires making good on those claims through demonstrable progress, not repackaging them in more upbeat language. The backlash, in his telling, reflects actual conditions—that AI’s benefits have not yet materialized at the scale or speed executives promised—rather than a failure of communication strategy or an excess of caution in public statements.