According to Anthropic, Fable 5 can autonomously perform many of the tasks involved in early-stage drug discovery like identifying protein binding sites, selecting appropriate computational design tools and navigating experimental setbacks without human intervention. Anthropic claims the model accelerated parts of the drug discovery process by approximately tenfold compared to human workflows.

In internal evaluations, the system generated promising drug candidates for 9/14 protein targets tested, with several candidates now progressing through further clinical assessment.

Beyond just drug design, Anthropic is also positioning Fable 5 as a research partner capable of generating novel biological hypotheses.

One example involved identifying a previously overlooked protein in Escherichia coli that may represent a new target for antimicrobial therapies. This hypothesis was subsequently validated through lab experiments.

The model also showed potential in gene therapy. Working with Dyno Therapeutics, Fable 5 explored challenges around adeno-associated virus vectors, which are widely used to deliver gene therapies. The model was tasked with predicting how genetic modifications could affect viral assembly, a complex problem that typically requires significant experimental effort.

However, Anthropic acknowledges that these capabilities introduce new risks, the same biological reasoning skills that could accelerate therapeutic innovation could also be misused. As a result, certain biological and chemical research queries are subject to additional safeguards and oversight.

Overall, Fable 5 offers pharma and biotech firms a glimpse into the next phase of AI adoption. The question is no longer whether AI can assist researchers, but how far it can participate in the scientific process itself.

“The AAV project demonstrates a promising ability to complete simple but important tasks in gene therapy research and development, but in the wrong hands, it could enable the design of dangerous viruses”

Anthropic