Anthropic CEO Amodei Refutes Pessimistic Criticism, Says AI Trust Crisis Stems from Unfulfilled Promises

As public discussions around AI risks and regulation continue to intensify, Anthropic's CEO Dario Amodei recently publicly refuted criticism that he was conveying an overly pessimistic picture. This response directly addressed comments made by investor Gavin Baker on the All-In podcast and social media platform X, who argued that Amodei's warnings about AI dangers have fueled public opposition to AI infrastructure such as data centers in the United States, and claimed that he has "lost" in the AI regulatory debate, suggesting that he should demonstrate a more positive stance as an industry leader.

Amodei denied that his messaging had a negative tone, emphasizing that his article "Machines with Love" aimed to "balance risks and benefits," and was intended to compensate for the industry's lack of depiction of the potential of technology to improve the world.

He acknowledged that the general public holds a negative view of AI and that this is a significant issue, but strongly denied that this sentiment primarily stems from his or other leaders' warnings about risks, instead attributing it to a "fundamental trust crisis" - long-standing public distrust of businesses and governments, with the AI backlash being the latest manifestation of this deeper contradiction.

In response to external criticism, Amodei admitted that the core issue that should be focused on is not the messaging but the fact that companies like Anthropic "have yet to fulfill their grand promises to benefit the world."

On regulatory positions, Amodei rejected Baker's binary choice of "widespread dissemination vs concentration of power," pointing out that regulation is often seen by outsiders as a tool to counter corporate power. He revealed that Anthropic's policy recommendations have consistently avoided clauses beneficial to leading large companies, instead supporting smaller competitors, because he firmly believes that AI tends to concentrate power structurally, and open-weight models cannot fully address this issue. He believes well-designed rules can simultaneously address AI's catastrophic risks, institutional constraints on leading companies' power, and leave room for open-weight models.