Biotechnology · global
AI-Designed Drug Delivers First Interim Human Signal, but the Real Test Is Just Beginning
Rentosertib’s Phase 2a trial moves AI drug development from demonstration platforms into the ward; it is not a declaration of victory, but a more testable starting point.
For years, the easiest thing to remember about AI drug development has been speed: finding targets faster, generating molecules faster, and advancing drug candidates faster. But the final judge of a drug has never been a model; it is the human body. During BIO 2026, the industry narrative took a clearer turn: rentosertib, an anti-fibrotic candidate discovered and designed through an AI process, has produced Phase 2a clinical trial results in patients with idiopathic pulmonary fibrosis, giving the question of whether AI can truly create drugs more concrete human evidence for discussion for the first time.
The protagonist of this result is rentosertib, developed by Insilico Medicine and formerly known as ISM001-055 or INS018_055. According to the randomized Phase 2a trial published in Nature Medicine, the study enrolled 71 patients with idiopathic pulmonary fibrosis and assigned them to placebo, 30 mg once daily, 30 mg twice daily, and 60 mg once daily groups, with treatment lasting 12 weeks. Idiopathic pulmonary fibrosis causes progressive scarring in the lungs and reduces gas-exchange capacity. Clinically, forced vital capacity, or FVC, is often used as one important indicator of changes in lung function.
The trial results showed that the proportions of treatment-emergent adverse events were broadly similar across groups, and treatment-related serious adverse events were also low and similar. In terms of efficacy signals, the mean change in FVC in the 60 mg once-daily group was an increase of 98.4 mL, while the placebo group showed a decrease of 20.3 mL. These numbers are not enough to directly declare the drug effective, but they are enough to move the discussion from whether the algorithm is elegant to whether the clinical signal is reproducible, whether there is a dose relationship, and whether it can hold up in a larger trial.
Rentosertib’s scientific path also makes it a representative case for AI drug discovery. A study published in Nature Biotechnology in 2024 stated that TNIK was identified by AI methods as an anti-fibrotic target, after which small-molecule TNIK inhibitors were generated and showed anti-fibrotic and anti-inflammatory activity in multiple in vivo fibrosis models. The study also reported a Phase 1 randomized, double-blind, placebo-controlled trial in 78 healthy participants, showing that safety, tolerability, and pharmacokinetic data supported further development; it took about 18 months from target discovery to preclinical candidate nomination.
The company’s press release added that the GENESIS-IPF trial was conducted at 22 centers in China and said the high-dose group showed time- and dose-related biomarker changes, including decreases in COL1A1, MMP10, and FAP, as well as an increase in IL-10. If these data can corroborate clinical endpoints, they will help clarify the mechanism of TNIK inhibition in pulmonary fibrosis. However, biomarkers remain supporting evidence, and larger and longer-term trials are ultimately still needed to confirm the actual effects on lung function, risk of worsening, and quality of life.
Background Context
The discussion at BIO 2026 was not only about a molecule, but also about industry competition. Tech Times interpreted this clinical progress in the context of the rise of Chinese biotech and political discussion in the United States related to the BIOSECURE Act, pointing out that Chinese companies have already taken the lead in obtaining symbolic results in clinical validation of AI drug development. This framing explains why the same set of trial data would be amplified into a geotechnological signal. But from a scientific perspective, national competition cannot replace clinical evidence, and policy anxiety will not lower the threshold required for drug approval.
Therefore, the most important meaning of rentosertib at present is that it brings AI drug development into a stricter and healthier stage. It is no longer merely demonstrating how to use models to find targets and design molecules, but is beginning to undergo the continuous tests of patients, dosing, safety, endpoint selection, and regulatory review. If the next step is to truly change the treatment landscape for idiopathic pulmonary fibrosis, larger-scale trials will still need to answer a simple but weighty question: whether this molecule born with the help of AI can steadily bring clinically meaningful improvement to more patients.