Anthropic's IPO filing softens its big claims about curing disease with AI

Anthropic filed paperwork for an initial public offering on Monday, and the document strikes a more cautious tone than the claims the company and its chief executive, Dario Amodei, have made about using artificial intelligence to transform biology.

The filing still says AI can “profoundly accelerate the rate of progress in biology and medicine” and compress “decades of scientific progress into years.” But it also admits to gaps the company has not had to address before. Anthropic’s teams “need first-hand experience doing scientific research and drug discovery,” the filing says, and moving into life sciences will require work “outside our historical experience as a software and research organization,” including handling hazardous materials, running laboratory equipment, and following a wide range of regulations.

That marks a shift from Amodei’s public statements. Two years ago, he wrote that he was open to the idea that AI systems could produce a thousand years of biological progress in five to ten years. In August, he posted on X that Anthropic was “ramping up its efforts very quickly in biology and medicine” and predicted AI would make it possible to cure most human diseases within five to ten years, promising a “deluge of AI-accelerated drugs.”

A quieter claim about the enzyme

In September, Anthropic promoted a discovery it said Claude had made: a “Crispr-like” enzyme, named for the gene-editing tool Crispr. The IPO filing describes that discovery more carefully, saying it “might have the potential to be developed into a novel biotechnology tool, perhaps for gene editing.” Researchers who spoke with WIRED previously said they were skeptical of the discovery’s value, and one scientist studying similar enzymes said he grew suspicious that Claude had used his own research; Anthropic has said neither its model nor its molecular biology team had access to that material.

The filing signals that Anthropic now treats biotech discovery as part of its core business, not only a tool it sells to other companies in the industry. That puts it on the same long timeline every other biotech company faces. Any AI-assisted discovery still needs testing in a test tube, then in animals, and eventually in people, before regulators will consider it. The filing spells this out, warning that “any drug candidates or other discoveries resulting from these efforts would face the lengthy, costly, and uncertain development and regulatory pathways characteristic of the life sciences industry, and may never generate revenue.”

SEC filings tend to be written more cautiously than public statements, since overstating claims to investors carries legal risk. Amodei, for his part, used his August post to argue for loosening regulations so AI-generated drugs could reach patients faster. The filing does not change Anthropic’s bet that its models will speed up biology; it changes how fast the company says that will happen, and how much still has to be built and proven before any of it reaches a patient. What to watch next is whether Anthropic’s path to an IPO surfaces more detail on its life-sciences unit, including any specific drug targets, lab investments, or regulatory steps tied to the enzyme discovery.