Anthropic CEO: AI must cure cancer to win public trust

Amodei's X post arrives at a moment of significant tension within the AI industry. The sector faces mounting public skepticism even as companies like OpenAI and Anthropic prepare for expected IPOs—a shift Amodei himself has noted, with both firms moving away from the cautionary tone that once dominated their messaging. Amodei's own track record includes repeated warnings about AI risks: in a June essay, he flagged "very real risks" posed by Anthropic's Mythos models to cybersecurity and critical infrastructure, and last year he predicted AI would eliminate half of all entry-level jobs. Yet he rejects the premise that his own warnings have driven public distrust, framing the skepticism instead as part of a broader decades-long erosion of corporate trust among Americans.

His proposal—that delivering genuine scientific breakthroughs could shift perception—has drawn pushback. OpenAI's Angel Brodin countered that the pharmaceutical industry routinely advances public health yet remains "one of the least trusted industries," and that companies will be judged not only on breakthroughs but on "pricing, access, lobbying, opacity, how the economic gains are distributed." Yann LeCun, Meta's former chief AI scientist, offered a different critique, arguing that Anthropic's decision to keep frontier models closed-source—rather than open-weight—contributes more to public distrust than anything else. LeCun has long advocated for open AI, saying "we need diverse AIs for the same reason we need a diverse press." Amodei, however, has stuck to Anthropic's closed-model stance, even as the company declined to sign a letter advocating for open-weight AI when Washington considered restrictions on some Chinese models last month.