UNC Lineberger gets $40M from OpenAI Foundation for cancer-vaccine AI data

UNC Lineberger Comprehensive Cancer Center said it received $40 million from the OpenAI Foundation to build clinical and biological datasets that can train AI models to pick better tumor targets for personalized cancer vaccines. The grant is part of the Foundation’s new Public Data for Health program, which funds open-access scientific datasets for life-science and disease research.

Same-day HEALTH primary that a major AI lab’s nonprofit parent is funding open biological datasets — not another chatbot feature — so models can learn from real tumor and immune data for personalized cancer vaccines. HEALTH desk priority: AI × cancer research with a named dollar figure and open-data intent.

UNC Lineberger Comprehensive Cancer Center, at the University of North Carolina at Chapel Hill, announced it received $40 million from the OpenAI Foundation. The UNC Health / UNC Lineberger newsroom post, dated September 15, 2026 and datelined Chapel Hill, N.C., is the filing event. These are UNC’s claims. This desk did not see the grant agreement.

Purpose, as UNC tells it: generate data that could make personalized cancer vaccines more effective. A personalized cancer vaccine is a shot tailored to a patient’s tumor markers so the immune system targets that cancer. The center says researchers will use AI to identify stronger cancer-cell targets and guide more precise vaccines. File that job as UNC’s. This desk did not review a vaccine trial result.

Named researchers, still UNC’s: immunologist Benjamin Vincent, MD, and computational biologist Alex Rubinsteyn, PhD. They are gathering clinical data; analyzing hundreds of de-identified tumor tissue and immune cells from three biobanks — stored tissue and cell samples for research; and comparing vaccine formulations in clinical trials for triple-negative breast cancer, a hard-to-treat subtype that lacks the usual hormone and HER2 targets. File the names, the three-biobank plan, and the triple-negative trial comparison as UNC’s. This desk did not count the samples.

Program claim, UNC plus a Foundation quote via UNC: the announcement coincides with the OpenAI Foundation launching its second science program, Public Data for Health — also called Data for Public Health in some coverage — aimed at high-quality open-access scientific datasets for life-science and disease research. An open-access dataset is research data other labs can reuse. File the program name and the open-data aim as UNC’s / the Foundation’s via UNC. This desk did not find a same-day openai.com or openaifoundation.org dated program page.

On-record Foundation voice, via UNC: Jacob Trefethen, Head of Life Sciences and Curing Diseases at the OpenAI Foundation, said AI progress in cancer treatment depends on having the right biological data, and that UNC’s work will generate data at scale and make findings broadly available. File the name, title, and that data-at-scale line as his, via UNC. His longer quotes stay in Sources.

Foundation identity, from UNC’s about blurb: OpenAI Foundation is described as a charitable organization and a separate nonprofit from OpenAI PBC that supports scientific discovery and public-benefit AI work. The same UNC lede also calls it the nonprofit parent that governs the for-profit company OpenAI Group PBC. File those identity lines as UNC’s. This desk is not inventing an endowment, an IPO date, or other grant amounts.

Plain English for the rest of the card: personalized cancer vaccine = a shot tailored to a patient’s tumor markers so the immune system targets that cancer. Biobank = stored tissue and cell samples for research. Open-access dataset = research data other labs can reuse. Triple-negative breast cancer = a hard-to-treat breast-cancer subtype that lacks common hormone and HER2 targets. Tumor antigen = a marker on cancer cells the immune system can learn to attack.

PRIMARY here: UNC Health / UNC Lineberger newsroom, 15 Sep 2026 — Tier A PRIMARY original record for the grant. Clear to file. Discovery only: MIT Technology Review’s same-day Data for Public Health wrap — Tier C corroboration, not the sole primary and not a source for extra grant amounts. The $40 million, the personalized-vaccine data purpose, Vincent and Rubinsteyn, the three-biobank / triple-negative trial plan, Public Data for Health, Trefethen via UNC, and the Foundation-as-separate-nonprofit identity are UNC-attributed. Foundation claims stay Foundation / company quotes via UNC. NOT claimed: IPO timing, a total Foundation endowment, other grant amounts beyond what UNC states, clinical efficacy results, that openai.com or openaifoundation.org posted a dated 15 Sep 2026 program page, that this desk reviewed patient data, a stock tip, or investment advice. Distinct from the already-filed gates-foundation-1b-ai-access, nvidia-chop-cardiac-monai, and powerful-medical-queen-of-hearts-fda-de-novo.