The change in tone
In the documentation filed on Monday for the stock exchange listing, Anthropic has dropped the enthusiastic tone used in the past to describe the role of artificial intelligence in biology. The company continues to maintain that its models can 'profoundly accelerate' progress in biology and medicine, compressing 'decades of scientific progress into years', but it is much more cautious about how that potential will translate into concrete results. According to WIRED's reconstruction, the document acknowledges that the company's teams 'need direct experience in scientific research and drug discovery'.
As the source reports, Anthropic writes that its life sciences activities will require 'engaging in activities outside our historical experience as a software and research organization'. These include managing potentially hazardous materials, using laboratory instruments and complying with numerous regulations. It is an admission that marks a departure from past statements by CEO Dario Amodei, who two years ago said he was 'open' to the idea that AI systems could produce a thousand years of biological progress in five or ten years.
The enzyme discovery and researchers' doubts
Another sign of caution concerns the discovery, announced in September, of a 'Crispr-like' enzyme identified by Claude. In the stock exchange document, the language has been toned down: the enzyme 'could have the potential to be developed into a new biotechnology tool, perhaps for gene editing'. The conditional wording replaces the more definitive statements in the previous communication.
WIRED recalls that some researchers had already expressed skepticism about the value and usefulness of that discovery, and that a scientist studying precisely those enzymes had suspected that Claude had used his research. The company stated that neither the model nor its molecular biology team would have had access to that material. The point remains controversial and is not resolved by the filed document.
The testing timelines and the regulatory framework
The listing application also clarifies a change in positioning: Anthropic no longer sees itself only as an external provider of AI tools for the biotech sector, but considers biological discovery a central part of its business model. This exposes it to the same difficulties as any biotechnology company. As the document states, any drug candidate or other discovery resulting from these efforts will have to face 'long, costly and uncertain development and regulatory pathways, characteristic of the life sciences sector, and may never generate revenue'.
The source stresses that filings with the U.S. Securities and Exchange Commission are typically more cautious than public statements, because of the legal risks of overstating prospects before potential investors. It remains confirmed, however, that the company believes in the revolutionary value of its technology. In his August post on X, Amodei had argued for the need to streamline rules for the future wave of AI-generated drugs.
What remains to be verified
The crux, according to WIRED, is that no AI-assisted discovery can skip stages: verification in a test tube is needed, then in animals and finally in humans. Even if artificial intelligence helped distinguish promising candidates from those destined to fail, years of testing would still be needed to demonstrate their safety and efficacy. The document does not indicate timelines, investments or expected results, and it is unclear what part of the business model will be entrusted to internal experimentation. The actual scope of the announced laboratory activities and the outcome of the listing process therefore remain to be established.
