OpenAI Foundation backs datasets to train medical AI

The OpenAI Foundation is funding a new Public Data for Health effort to create high-quality scientific datasets for medical AI. One supported project comes from Ruxandra Teslo, a clinical trials policy analyst, who proposed acquiring regulatory filings, manufacturing strategies, safety data, and other trade-secret-like materials from failed biotech companies through bankruptcy proceedings.

The foundation also announced support for a University of North Carolina, Chapel Hill program collecting data on novel cancer vaccines and for OpenAdmet, which runs competitions to predict drug effects. Morgan Levine, a former vice president for computation at Altos Labs, said data has become the biggest bottleneck in applying AI to biology.

Teslo’s biotech archive project will be pursued by 1Day Sooner, an advocacy group for clinical trial volunteers. The group already holds three datasets, including two donated by Lumen Bioscience, and has made unsuccessful bids for other drug company files this year. The targeted records, known as common technical documents, include regulator-company exchanges and detailed scientific and medical measurements.

The push comes as the OpenAI Foundation expands its grantmaking while holding a 26% equity stake in OpenAI. Wider concerns over AI risk remain prominent, with some insiders putting the chance of human extinction within the next decade at 10% or more.