Anthropic CEO Dario Amodei has rejected claims that his warnings about artificial intelligence are driving public skepticism, arguing that the backlash reflects a deeper crisis of trust in companies, governments and the technology industry.
The comments followed investor Gavin Baker’s argument that Amodei’s warnings about AI risks had contributed to opposition to data centers and broader resistance to the technology in the U.S. Baker urged Amodei to take a more positive stance as the head of a major AI company.
Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram
Amodei disagreed, saying his writing has been roughly balanced between AI’s risks and benefits. He pointed to his essay “Machines of Loving Grace,” which explored how AI could transform society for the better.
Still, he acknowledged that public opinion toward AI is negative and called it “a big problem,” while rejecting the idea that AI executives warning about risks are primarily responsible.
“I think it is fundamentally a crisis of trust,” Amodei said, arguing that many people already distrust companies, governments and the technology industry.
AI Companies Need To Deliver
For Amodei, the strongest criticism of AI companies is not their messaging but the gap between their promises and results. He said the industry needs to demonstrate tangible benefits rather than simply promote them, arguing that actually curing diseases such as cancer would do more to change public opinion than promising that AI might one day do so.
Amodei Defends AI Regulation
Amodei also rejected the idea that regulation necessarily concentrates AI power among the largest companies, calling it a “false choice.” He said carefully designed rules could constrain corporate power while giving smaller competitors room to grow.
Anthropic has supported measures including transparency requirements for large AI companies. Amodei said its proposals aim to slow down frontier AI companies while benefiting smaller competitors.
He also argued that AI is “structurally” prone to concentrating power. Open-weight models can distribute some of that power, but access to computing resources and advanced chips remains concentrated.
In his view, effective regulation should address AI’s cybersecurity, biological and alignment risks while limiting the power of major AI companies and preserving room for open-weight models.
Ultimately, Amodei suggested that rebuilding trust will depend less on how the industry talks about AI and more on whether it can deliver meaningful benefits while addressing its risks.
