Anthropic CEO says the way for AI to win over the public is to cure cancer

In a rare post on X, Anthropic CEO Dario Amodei acknowledged public mistrust of AI and argued that the industry can only change that by delivering concrete scientific breakthroughs. He wrote that saying AI will cure cancer has become a cliché many view as deceptive, adding that "The thing that will work is actually curing cancer." He also admitted that AI companies, including Anthropic, haven't yet delivered on their big promises to benefit the world.

The comments come amid growing public backlash over AI's real-world effects: companies are building large data centers in communities, training models on copyrighted online data, and disrupting workforces. A recent Pew Research Center study found that roughly half of Americans feel more concerned than excited about AI's increasing presence in daily life.

Amodei has frequently warned about AI risks in podcasts, essays, and blog posts — citing "very real risks" from the company's newest models to cybersecurity, finance, critical infrastructure, and national security, and predicting that AI would eliminate half of all entry-level jobs. Some peers, like Google DeepMind's Demis Hassabis, have criticized such doomsaying as "way too certain." As OpenAI and Anthropic prepare for possible IPOs, they've pivoted from doomerism to boomerism.

Amodei pushed back on the notion that his negative messaging is the cause of public distrust, saying the skepticism reflects broader American distrust of corporations that predates AI. He pointed to his 2024 essay "Machines of Loving Grace" as an attempt to paint an inspiring picture of AI's potential. He also noted that Anthropic's decision not to open-source its frontier models — made out of caution — has contributed to public wariness; the company drew criticism last month for being the only major AI firm to keep its models closed-source.