AI models need more data about biology, and OpenAI is paying to create it

The OpenAI Foundation has launched a new initiative, Public Data for Health, aimed at addressing the critical data shortage in medical AI. Recognizing that high-quality datasets are the primary bottleneck for biological breakthroughs, the foundation is funding projects to generate and curate scientific information. Among the initial recipients is a $500,000 grant to 1Day Sooner, which will explore sourcing data from bankrupt biotech companies—a treasure trove of regulatory filings and safety data often lost to history. Additionally, the foundation is providing $40 million to support cancer vaccine research and drug effect prediction programs. As the OpenAI Foundation manages a massive endowment tied to OpenAI's equity, it aims to distribute $1 billion in grants by year-end. This effort reflects a strategic shift toward fostering external research to ensure AI development benefits humanity, even as the industry grapples with broader concerns regarding AI safety and potential existential risks.