AI Drugs: 129 Candidates Tested on Humans, No Approval Yet
129 AI-developed drug candidates had entered human testing by March 2026, five in phase III. None of the active ingredients has been approved yet.
Here are the numbers on drugs and artificial intelligence: by March 2026, 129 therapeutic candidates developed with the help of AI had started testing in humans. A therapeutic candidate is a molecule or treatment still under investigation, not an approved drug.1 Five of these candidates have reached phase III or have already completed it. The information available so far does not show that they perform better, clinically, than classic drug research methods.1
Where approval stands
In the United States and the European Union, no new active ingredient designed with AI has so far received permission to be sold.1 One candidate with a documented contribution from artificial intelligence, zasocitinib, is nonetheless under review by the FDA, the US drug agency, and by the EMA, the European one.1 The product is designed for psoriasis and has completed the decisive phase III trials. In the United States the application is following a priority review pathway, and the agency’s response is expected within the first three months of 2027.1
A trial involving Italy
A trial is also underway on an antibody designed with machine learning techniques, meaning automated learning, against severe asthma. The study also involves Italy.1
What changes
The picture is that of many candidates in human testing: five have reached the final phase or beyond, and no active ingredient has yet been authorized.1
This article was written by artificial intelligence from the sources cited, without human review. Automatic newsroom by Redazia.
