AI, Anti-Aging, and Protecting What AI Creates
For decades, “anti-aging” meant creams and supplements with little proof behind them. In 2026, AI-designed drugs are in late-stage trials and the first cellular rejuvenation therapy has reached a patient. Aging is a hard target because it is not one disease but a slow breakdown across thousands of genes and proteins at once. AI can find patterns in that data that humans cannot, and AI-built “aging clocks” can estimate biological age from blood, so researchers can test ideas in months instead of decades.
Insilico Medicine shows AI working from start to finish. Its platform identified TNIK as a target linked to both aging and fibrosis, and its generative chemistry engine designed the drug, rentosertib. On September 10, Insilico dosed the first patient in a Phase III trial for idiopathic pulmonary fibrosis, an age-related lung disease. Days earlier, a reanalysis of mid-stage trial samples found that all six independent aging clocks rated treated patients as biologically younger. The authors caution that this does not yet prove body-wide slowing of aging rather than lung healing.
Retro Biosciences used OpenAI’s GPT-4b micro to redesign two Yamanaka factors, the proteins that return cells to a stem-cell-like state, and the new versions showed over 50-fold increases in reprogramming marker expression. That work is preclinical, but Retro’s autophagy pill RTR242 is in a Phase 1 trial in healthy volunteers in Australia.
BioAge Labs mines data from long-lived people. Its platform picked NLRP3 because lower NLRP3 activity is associated with greater longevity, and its oral inhibitor BGE-102 cut the inflammation marker hsCRP by a median of at least 85% in Phase 1.
Company | Candidate | Indication | Stage (Sept 2026) | AI role |
Insilico Medicine | Rentosertib | Idiopathic pulmonary fibrosis | Phase III | Target and molecule design |
BioAge Labs | BGE-102 | Cardiovascular risk; diabetic macular edema | Phase 2; Phase 1b/2a | Target found in human aging cohorts |
Rubedo Life Sciences | RLS-1496 (topical) | Psoriasis, dermatitis, skin aging | Phase 1 complete, positive preliminary results | ALEMBIC discovery platform |
Retro Biosciences | RTR242 | Alzheimer's-type decline | Phase 1 | Protein engineering platform |
NewLimit | Liver mRNA therapy | Fatty liver disease | Preclinical; trial planned 2027 | Machine learning with genomic screens |
Nearly every company enters the clinic through a specific disease, because regulators do not treat aging as one. Any aging benefit is measured alongside the main result.
The inventor must be human. The USPTO’s November 2025 guidance confirms that only natural persons can be inventors and treats AI as a tool, like lab equipment. India’s Patent Office refused the AI-inventor DABUS application in April 2026, and an appeal is pending before the Delhi High Court. The UK Supreme Court reached the same result in Thaler (2023).
Patenting the AI itself is getting easier in some places. In Emotional Perception [2026] UKSC 3, the UK Supreme Court dropped the Aerotel test and aligned with the European Patent Office. India’s CRI Guidelines 2025 allow AI inventions only where they show a real-world technical effect.
A four-layer strategy works best:
1. Patent the product. The molecule, protein, or mRNA construct is the core asset. In India, claim compounds and compositions, since Section 3(i) bars treatment methods and Section 3(d) limits new forms of known substances.
2. Document human contribution. Keep records of the targets scientists chose, the goals they set, and the outputs they selected and tested.
3. Keep the platform secret. Training data and model weights are often better protected as trade secrets. OpenAI is not releasing GPT-4b micro publicly.
4. Watch your tools. AI speeds up prior-art searches and freedom-to-operate checks, but pasting unfiled inventions into consumer AI tools can put secrecy and novelty at risk.
No drug has yet been proven to slow human aging, and most early-stage drugs fail. Still, AI is now finding targets, designing molecules, and measuring results in ways it could not a few years ago. The companies that get both the science and the IP right will lead the field.