Policy analyst Ruxandra Teslo pitched using failed biotech firms’ filings and safety data to train medical AI.

The OpenAI Foundation, the nonprofit parent of OpenAI, said this week it will fund an effort to build high-quality scientific datasets about biology. The money backs an idea from policy analyst Ruxandra Teslo, who proposed sourcing data from failed biotech companies.

Teslo argued last year that bidding at bankruptcy proceedings could unlock detailed regulatory filings, manufacturing strategies, and safety data from defunct biotech firms. She described that material as biotech’s lost archive, a trove that medical AI systems currently lack.

AI models need far more biological data to make progress on curing disease. Operators building clinical or drug discovery tools often hit the same wall: companies that fail rarely publish negative results, so models train on thin evidence.

What comes next is whether the funded datasets improve model performance on real biology tasks. Watch how the foundation structures data access, licensing, and provenance, and whether other labs can reuse the archive for their own training runs.

What matters

• The OpenAI Foundation will fund creation of high-quality scientific datasets for medical AI.
• Failed biotech firms hold regulatory filings and safety data that few AI teams can access today.
• Watch whether the funded datasets sharpen biology predictions and reach other research labs.

Why it matters

Watch whether the funded datasets sharpen biology predictions and reach other research labs.

This GenAI News article was prepared in original wording using reporting and materials published by MIT Technology Review AI. Source reference: https://www.technologyreview.com/2026/09/16/1144205/the-download-ai-trillion-dollar-build-openai-biological-data/.

Drafted by the GenAI News review pipeline.