Anthropic says Claude proposed 35,000 antibodies for a food-allergy drug
Anthropic says its AI produced a huge batch of candidate antibodies with a biotech partner. The work is early, but it shows AI moving into the lab.
The short version
- Claude produced about 35,000 antibody designs in roughly a month, with Divergence Labs.
- Three designs seemed to pull IgE loose better than a standard medication.
- Next step: try combining the designs into a single drug.
- Claude also helped cut human review time by 89% on documents at Enveda.
- The projects are early and mostly preclinical, so no human trials yet.
Why it matters
AI is being used on the slow, expensive front end of drug discovery, not just on emails and images. If it can offer scientists many more candidates to test, the search for new medicines could speed up. But a long list of ideas is not a treatment, and the real tests are still ahead.
A big batch of candidates
Anthropic says its Claude model came up with about 35,000 proposed antibodies in roughly a month. The target was a potential food-allergy treatment, and the work was done with a biotech company called Divergence Labs, as KUTV reported.
Antibodies are proteins the immune system uses to latch onto things. A "design" here is a proposed protein shape, not a finished medicine.
What the early tests showed
The researchers aimed at IgE, a molecule that plays a part in allergic reactions. In early tests, they found designs that could attach to IgE or knock it loose from where it sits.
Anthropic said three designs seemed to dislodge IgE better than a standard medication. The teams now plan to try merging the designs into one drug.
Beyond allergies
Anthropic said it is working with five other organizations that are developing medicines. In one project, Claude helped researchers shorten the genetic instructions for a cancer treatment. In lab tests, immune cells given the shorter version made about 10 times as many cancer-targeting receptors.
The company also described plainer jobs: checking manufacturing records and drafting papers for regulators. At Enveda, it said, Claude cut human review time on one set of documents by 89%.
How far this really is
Everything described is early, and most of it is preclinical. That means the treatments have not been given to people in trials. These are also Anthropic's own claims about its own tool.
The big number shows how quickly AI can generate options. It does not show how many of those options will work in the body. Anthropic says it plans to grow this work.
What to watch
- Whether the combined allergy drug works in lab tests.
- Whether any of this work reaches human trials.
- Whether outside scientists confirm the results.
Try it
When you see a large count in an AI claim, ask one question: how many of these were actually tested, and what happened?
Sources and links
- ReportingKUTV: the article we sourced
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.
Listen to the AI Wire headlines
Listen to the AI WireListen to our daily AI shows every morning: on your commute, in the shower, wherever you are
Listen to the daily showsFirst reported by KUTV · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.