OpenAI foundation funds new biological datasets to boost medical AI
The OpenAI Foundation has launched Data for Public Health, an initiative to fund creation of high-quality scientific datasets for medical AI. Initial grants include $40 million for cancer vaccine data collection at the University of North Carolina and $500,000 to pursue a proposal to acquire data from bankrupt biotech companies. The effort aims to address the data bottleneck that limits AI breakthroughs in biology and medicine.
The OpenAI Foundation's new grant program targets what researchers describe as a critical shortage of biological training data. Initial funding includes $40 million for cancer vaccine data collection at the University of North Carolina and $500,000 to pursue Ruxandra Teslo's proposal to acquire regulatory filings and safety data from bankrupt biotech firms through bankruptcy proceedings. The foundation also supports OpenAdmet, which runs drug-effect prediction competitions.
The foundation holds a 26% equity stake in OpenAI, potentially worth $250 billion as the company plans a $1 trillion IPO. That would make it the world's wealthiest charity, surpassing the Gates Foundation's roughly $180 billion. Still, the organization is building its grantmaking capacity, having awarded only $100 million so far, and its efforts coincide with industry leaders calling for slower AI development amid extinction risk concerns.
This initiative could reshape medical research by making proprietary drug data publicly available, potentially accelerating AI-driven discoveries in disease treatment and vaccine development. Patients may benefit from faster breakthroughs, while biotech companies could face new competitive pressures if their failed ventures' data becomes training material. The foundation's enormous projected wealth means its funding priorities could significantly influence which medical research areas advance, though the long-term societal impact will depend on whether these datasets actually produce usable AI insights.