AI-Designed Lung Fibrosis Drug Enters Phase III on Small Early Data

Idiopathic pulmonary fibrosis is an age-related scarring of the lungs. The approved drugs slow it but do not reverse it. A drug built from start to finish with generative AI is now in the large trial that will show whether it does better.

Insilico Medicine announced on September 10, 2026 that it dosed the first patient in GENESIS-IPF-3, which the company calls the world's first Phase III trial of a generative-AI-driven drug. The drug targets idiopathic pulmonary fibrosis (IPF).

The company's release confirms the trial details. GENESIS-IPF-3 (NCT07687459 / CTR20262475) dosed its first patient on September 9, 2026 at Peking Union Medical College Hospital. It is a 52-week, randomized, double-blind, placebo-controlled study of 320 participants at 47 centers in China. The primary endpoint is the annual rate of FVC decline over 52 weeks; a key secondary endpoint is time to first diseaseprogression event. The claim to be first comes from the company and is not independently confirmed.

The drug's earlier record is solid and published. Rentosertib, formerly ISM001-055, blocks a protein called TNIK. Both the target and the molecule came from Insilico's generative AI platform.

The Phase 2a trial appeared in Nature Medicine in June 2025. It was double-blind, placebo-controlled and run at 22 sites in China:

• 71 patients were randomized to one of three dose arms or placebo for 12 weeks. • The main goal was safety. Side-effect rates were similar across arms, from about 71% on placebo to 83% on the higher doses. • Stopping treatment was most often due to liver toxicity or diarrhea. • At the highest dose, lung capacity (forced vital capacity) rose by an average of 98 mL, versus a fall of 20 mL on placebo. That lung-function gain is encouraging because IPF usually means steady decline. But it came from about 18 patients per arm over three months, and the confidence interval was wide, roughly 11 to 186 mL.

A 2026 network meta-analysis in BMC Pulmonary Medicine, pooling 35 trial reports, placed rentosertib among the agents most likely to preserve lung function. The authors rated that evidence low certainty, built mostly on indirect comparisons.

There are two reasons to watch this trial. The first is patients. IPF mostly affects people over 60, and median survival after diagnosis is measured in a few years. A drug that stabilizes or improves lung function would matter enormously.

The second is the AI question. Many companies claim AI speeds up drug discovery. A positive Phase III would be among the first hard tests of that claim at the stage where most drugs fail.

There is also a longevity thread. In September 2026, Insilico and academic coauthors reported in Nature Biotechnology that six proteomic aging clocks predicted lower biological age in treated patients from the Phase 2a trial. The authors themselves note that clocks cannot separate an aging effect from a disease effect. Insilico's founder is the first author, so we treat this as an interesting, company-led signal.

• Size. Twelve weeks and 71 patients cannot predict long-term survival or exacerbations. • Generalizability. The Phase 2a was run in China, and the Phase III is also enrolling only in China. Whether the effect holds in broader populations will remain open after this trial. • The aging claim. Lower clock scores are a biomarker, not proof of slower aging. • AI credit. A win would show the molecule works. It would not by itself prove AI was faster or cheaper than conventional discovery.

• If you have IPF or another fibrotic lung disease, ask your pulmonologist about clinical trials of new agents. Note that GENESIS-IPF-3 enrolls only in China, so most readers will not have access to it. • Report a new dry cough or breathlessness on exertion that lasts more than a few weeks, especially after 60. Early diagnosis opens more options. • Do not seek this drug outside a trial. Its long-term safety, including liver effects, is still being studied.

• Whether the 52-week primary endpoint, the annual rate of FVC decline, shows a clear benefit over placebo. • Whether the key secondary endpoint, time to first disease-progression event, points the same way, and whether exacerbations, hospitalizations and survival are reported. • Liver safety signals as more patients are treated for longer. • Whether any trial outside China tests the drug in broader populations.

Same Immune Effort, More Survivors: Why Uneven Contact With Target Cells Could Matter AI-Designed Lung Fibrosis Drug Enters Phase III on Small Early Data Could Stiffening Gut Tissue Block Protective Antibodies as We Age? A Long-Shot Idea, Rated 25 Percent Could Methionine Restriction Backfire on the Liver After Menopause? A Testable Idea About Choline, Rated 60 Percent