Anthropic and CEO Dario Amodei have spent years promising that its AI models will transform biology and drug discovery. Monday's IPO filing tells a more measured story.

The company acknowledged it lacks hands-on experience in physical experimentation and flagged the risks of drug development. Two years ago Amodei said he was "open" to AI producing a thousand years of biology progress in five to 10 years; in August he predicted a "deluge of AI-accelerated drugs" and cures for most human diseases within a decade.

The filing still says AI can "profoundly accelerate" progress and compress "decades of scientific progress into years," but is guarded about converting potential into results. Anthropic says its teams "need first-hand experience doing scientific research and drug discovery," work that falls "outside our historical experience as a software and research organization"-handling hazardous materials, running lab equipment, and meeting regulations.

Language around its Claude-discovered "Crispr-like" enzyme was softened too: it "might have the potential to be developed into a novel biotechnology tool, perhaps for gene editing." Researchers had told WIRED they doubted the enzyme's value, and one scientist studying those enzymes suspected Claude used his research; the company said neither its model nor its molecular biology team had access to the material.

The filing shows Anthropic now treats biotech discovery as core to its business, not just a customer market-and it concedes any drug candidates face "lengthy, costly, and uncertain" development and may "never generate revenue." SEC filings are typically more cautious than executive statements, but the message is clear: proving worth in trials can't be sped up by adding computers.

Source: Wired science