Anthropic says public opposition to AI stems from a deep-rooted crisis of trust and delayed delivery on major promises

Dario Amodei, chief executive of Anthropic, argues that public pushback against artificial intelligence does not stem from warnings issued by tech leaders about model risks, but is rooted in an accumulated crisis of trust between the public, corporations, and governments. Amodei made his remarks in a series of posts responding to investor Gavin Baker, who criticised on the All-In podcast and the X platform what he described as excessive negative messaging from industry leaders, arguing that risk warnings have fuelled attacks on data centres and weakened the industry's standing in regulatory debates, while calling on Amodei to adopt a more positive tone and defend his sector as his company moves toward becoming one of the most prominent firms in the world.

Amodei rejected the premise that his messaging is disproportionately negative, explaining that his writings divide attention equally between benefits and risks. He pointed to his previous essay on positive upside, written to address the lack of an inspiring vision for how technology can improve the world, while acknowledging that general sentiment toward artificial intelligence is negative and presents a genuine problem. Yet he stressed that this negative perception did not arise from warnings by industry leaders, but because ordinary people distrust tech companies and governments, expecting them to find new ways to harm their interests, adding that this crisis has existed for decades and the current backlash is merely its latest chapter.

The head of Anthropic considered the most accurate criticism levelled at AI companies, including his own, to be the failure so far to deliver on grand promises of tangible benefits to the world.Repeated promises to cure intractable diseases such as cancer have become worn tropes that lack inspirational power, as public convictions will only shift when actual cures are delivered in practice, emphasising that failing to deliver these benefits is the real shortcoming companies should be held accountable for, rather than focusing on marketing campaigns and media messaging.

On regulation and policy, Amodei criticised the common association in Silicon Valley between regulatory rules and regulatory capture, calling that view overly reductive. He explained that the public outside those circles views regulation as a means to rein in the influence of large corporations and protect the public interest.Balanced legislation aims to slow down and constrain frontier AI companies while giving an advantage to smaller competitors, stemming from the structural nature of artificial intelligence, which naturally pushes it toward the concentration of power and control.

Amodei concluded his stance by noting that open-weights models, while partially helpful, are not a sufficient solution to prevent the concentration of power because they shift control toward entities possessing the largest compute infrastructure and chip supplies. He maintained that crafting precise regulations can address cyber, biological, and alignment risks, while placing institutional constraints on the leverage of leading firms and leaving sufficient room for open models to address their specific associated risks.