OpenAI Foundation: The Why and How of Its Life Science Research Funding Agenda

Key Takeaways

• The OpenAI Foundation is investing billions of dollars in AI-driven scientific research, with life sciences, drug discovery, and curing diseases among its central priorities.

• The OpenAI Foundation’s life sciences funding strategy targets three major opportunities: accelerating Alzheimer’s disease research, expanding access to high-quality biological data sets, and developing treatments for high-burden diseases.

• AI could transform drug discovery and development by integrating complex biological data, identifying novel therapeutic targets, improving molecular design, and reducing the time and cost required to develop new medicines.

• OpenAI Foundation grants are supporting AI-enabled Alzheimer’s research, open biomedical datasets, biomarker discovery, and initiatives designed to improve the efficiency and success rates of pharmaceutical R&D.

• OpenAI Group is advancing AI tools for life sciences research, including GPT-4b micro for protein engineering and GPT-Rosalind for biological reasoning, experimental planning, and drug discovery workflows.

OpenAI began in 2015, initially founded as a nonprofit with the mission to make sure artificial general intelligence (AGI) “benefits all of humanity.”¹ In 2019, the organization created a for-profit subsidiary “governed and controlled by the nonprofit” to support deployment of its developed technologies at scale. In October 2025, recapitalization and restructuring resulted in the formation of the OpenAI Foundation as the nonprofit entity and OpenAI Group PBC as a for-profit public benefit corporation, both still having the same mission. As a public benefit corporation (PBC), OpenAI Group is required to “advance its stated mission and consider the broader interests of all stakeholders, ensuring the company's mission and commercial success advance together.”²

The OpenAI Foundation held a 26% equity stake in OpenAI Group PBC at the closing of the recapitalization, worth approximately $130 billion based on OpenAI Group’s valuation at the time.² That stake will increase notably if OpenAI Group’s valuation rises by more than a factor of 10 after 15 years. This approach links the resources of OpenAI Foundation to the success of OpenAI Group, makes it the single largest long-term beneficiary, and according to the organization, ensures the mission remains the focus of both entities.¹ The recapitalization and restructuring moves were made following a year of consultations with the offices of the Attorneys General of California and Delaware.

Together, the OpenAI Foundation and OpenAI Group will “work in concert to advance solutions to hard problems and opportunities posed by AI progress,” including “make[ing] intelligence a tool that everyone can benefit from, building safe and aligned systems, turbocharging scientific discovery, and strengthening global cooperation and resilience.”¹ Initially, the focus will be on two main areas: life sciences and AI resilience. Efforts will be directed toward using AI to enable “faster diagnostics, better treatments, and cures” and “maximiz[e] AI’s benefits and minimiz[e] its risks.” The foundation is also interested in helping communities manage the impacts that AI will have on jobs and the economy and investing in community-based organizations and initiatives that “focus on helping people understand AI, benefit from its capabilities, and adapt to the changes it brings.”³

The biopharmaceutical industry has helped extend the human lifespan and quality of life by developing medicines that treat and in some cases cure diseases. Over time, however, drug development has become more challenging, with the innovation rate dropping while both costs and development timelines steadily increase, and conditions such as diabetes, heart disease, neurological disorders, and many forms of cancer still lack effective therapies. This decline in productivity has occurred despite reduced U.S. Food and Drug Administration (FDA) review times and dramatic technology advances (e.g., combinatorial chemistry, DNA sequencing, analytical sensitivity, omics, bioinformatics).

OpenAI believes that AI can accelerate drug development, save both time and money, and bring novel treatments to more patients sooner.⁴ That is because AI can help scientists overcome the increasingly siloed nature of research today. Unlike historically famous scientists, such as Pasteur and Darwin, who worked across many fields, general topics, such as natural history, are today broken into a multitude of specialist areas with little cross-talk. In addition, there are large amounts of diverse data types available that AI can mine for hidden insights.

AI is already being used to improve protein and molecular structure determination, the design of novel gene editors, the development of self-driving labs, automated data analysis for target/biomarker discovery, optimal candidate selection, and patient selection and disease assessment in clinical trials, among other applications.

“With AI, the life sciences are moving back toward a fusion of disciplines that the burden of knowledge has driven apart,” say OpenAI’s Abhishaike Mahajan and Jacob Trefethen. “Genomics turned the genetic code into a digital stream of information. CRISPR and its successors, including base editing, prime editing, and epigenetic editing, made that code writable and programmable with growing precision. AI systems trained on massive biological datasets began to interpret the code’s complexity at a scale no specialist could match. Robotic automation is closing the loop between computation and wet-lab validation. Together, these advances create a compounding cycle: reading genomes, computing predictions, writing changes, and validating them in automated labs.”

Maximizing the potential of AI, meanwhile, will require approaching AI in the life sciences as a “shared national research resource” and building the infrastructure needed to turn AI-generated results into real world–validated products, which will require “investing in AI infrastructure hubs, streamlining permitting for high-capacity compute, and expanding the experimental and trial capacity,” alongside efforts to “support upskilling at all levels of education, expand certification programs, apprenticeships, and community-based AI training so the benefits of AI-enabled science are shared broadly.”

AI is already being used for many applications within the life science sector, from drug discovery to image analysis, development of predictive models for improved drug formulation and manufacturing processes, and much more. The successes realized to date underscore the significant potential for AI to positively impact the life sciences and health.⁵

In addition to its initial commitment to invest $25 billion in curing diseases and AI resilience, the foundation announced in 2026 that it would be investing at least an additional $1 billion across life sciences and curing diseases, jobs and economic impact, AI resilience, and community programs.³ The goal in all areas is to “enable the use of AI to find solutions to humanity’s hardest problems, transform what people are capable of, and deliver real benefits in people’s lives — while working hard with partners to be ready for new challenges, and to help make society resilient, as AI advances.”

To that end, the Life Sciences & Curing Diseases group within the OpenAI Foundation, headed by Jacob Trefethen, will be focusing on finding effective treatments for Alzheimer’s disease, increasing access to public data for drug development and health-related applications, and discovering medicines for high-burden diseases (“any disease that leads to over half a million deaths around the world each year or similarly massive ill health.”⁴ The organization believes that “AI has enormous potential to speed up scientific and medical progress to save and improve lives.”³

In Alzheimer’s research, AI is anticipated to help uncover insights within vast amounts of diverse data that human researchers cannot find. Initial work will include disease pathway mapping, biomarker identification and detection, personalization of treatments, and drug repurposing. To increase data access, the Foundation will support researchers seeking to “expand open, high-quality datasets” and “where appropriate, help responsibly open previously closed ones.”3 With respect to “high-mortality and high-burden diseases,” the focus will be on lowering the risk and cost of drug development.

In April 2026, the OpenAI Foundation announced it would be awarding $100 million in grants to six research institutions leveraging AI as a means to increase the disease knowledge base and accelerate drug development.⁶ Alzheimer’s is a natural focus for the foundation as it affects a growing percentage of the global population and is a highly complex disease for which AI capabilities should be highly beneficial. While novel therapies have been introduced in recent years to treat heart disease, infectious diseases, and several cancers, little progress has been made in the area of neurodegenerative diseases.

Alzheimer’s poses a big challenge to drug developers because numerous factors contribute to its development, including heredity and environmental conditions that can influence protein misfolding, inflammation, synaptic dysfunction, and other biological functions. The ability of AI to reason across different types of data makes it ideally suited to uncover previously unknown interactions of these factors and reveal novel biomarkers of disease and drug targets that will lead to more effective treatments.

Working with external experts, the OpenAI Foundation has established a five-pronged initial strategy for tackling this significant problem that will complement efforts already underway. These include:

• Creating a map of all possible causal factors that depicts their interactions and reveals the most promising points for therapeutic intervention as a function of individual disease attributes;

• Designing new drugs with AI biology tools and validating them in the lab;

• Working with various nonprofits to develop new open Alzheimer’s disease–relevant datasets and expand existing data sets in a way that enables responsible access;

• Identifying new disease biomarkers that enable earlier diagnosis and facilitate clinical trials; and

• Investigating existing drugs with limited off-patent data suggesting potential effectiveness.

OpenAI is collaborating with partners such as the Arc Institute, the Institute for Protein Design⁠⁠, the Mass General Brigham Neuroscience Institute, EVE Bio, and the University of California, San Francisco (UCSF), in these efforts. The initial strategy will be modified as initial results are obtained and further feedback is received from the wider Alzheimer’s research community.

In September 2026, OpenAI launched its second Life Sciences and Curing Diseases program, this one focused on Public Data for Health and specifically “the creation and preservation of high-quality scientific datasets made broadly available to researchers.”⁷ Underlying these efforts is the belief that applying AI to the analysis of more high-quality data will lead to novel treatments for preventing and treating diseases. Grants totally over $125 million are being awarded to nonprofits and universities working in many areas of life science research.

The foundation’s strategy is to support the development of data sets that are connected and contain currently scarce data, as well as data obtained from direct measurement of biological attributes, and to make these data sets accessible while ensuring privacy where appropriate. Funding will be going to research teams that are developing rich datasets covering biological aspects at different scales — such as at the single-molecule and tissue levels — and demonstrating how the different data interact. Groups seeking to collect or preserve valuable data before it disappears or make existing private data sets broadly accessible while still respecting privacy and individual consent are also receiving support from the foundation. Finally, the organization is helping researchers develop new tools that enable direct measurement of important “biological truths” (new data) not previously studied.

All researchers receiving funding support from the foundation must also have plans for sharing their results on a global basis, again in a manner that maintains privacy and consent when human/patient data is involved, such as through publishing their analyses via preprints, sharing data throughout from the start to the end of the project, and if not possible to include in public data sets, then working directly with interested users.

The first grantees include the OpenADMET team at UCSF, which is developing open data sets, benchmarks, and blinded competitions to test whether AI models can be trained to accurately predict the absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties of drug candidates, helping to eliminate candidates unlikely to be safe and effective early in development and thus increase the development success rate. The CTD Commons group is seeking to collect the data reported in common technical documents for drug candidates that have failed to receive regulatory approval or whose programs have been cancelled and make them openly available. They will also use machine learning systems to discover patterns associated with candidates that fail or succeed, information that can be used to improve drug development, clinical trial design, and regulatory development strategies. The Initiative for Generative Immunotherapy at the University of North Carolina (UNC) has received a grant to support the direct measurement of tumor cell surface proteins and patient T cell responses, information that will be used to identify better targets for next-generation personalized cancer vaccines.

To aid its mission, both the OpenAI Foundation and the for-profit OpenAI Group have focused on assisting life science researchers, the latter through the development of AI tools. Two primary examples: GPT 4b micro⁸ and GPT Rosalind.^9,10

GPT 4b micro is a miniature version of the multimodal AI model GPT 4o (which processes text, audio, images, and video in a single neural network) specifically designed to enable protein engineering.⁸ GPT 4b micro leverages GPT 4o’s existing knowledge and was also trained using a data set comprising largely protein sequences, as well as biological text and tokenized 3D structure data that most protein language models do not contain. This information includes “textual descriptions, co-evolutionary homologous sequences, and groups of proteins that are known to interact.” As a result, GPT 4b micro can generate sequences with specific desired properties, including disordered regions as well as structured proteins.

The capabilities of GPT 4b micro were demonstrated by researchers at Retro Biosciences, who used the AI model to design improved variants of the Yamanaka factors SOX2 and KLF4, proteins used to generate induced pluripotent stem cells (iPSCs). The designed proteins were shown to exhibit more than 50-fold higher expression of stem cell reprogramming markers than wild-type controls and enhanced DNA damage repair abilities. The initial findings obtained in 2025 were subsequently replicated across multiple donors, cell types (mesenchymal stromal cells (MSCs)), and delivery methods (viral vectors and messenger RNA).

GPT-Rosalind is OpenAI’s purpose-built AI model for research applications across biology, drug discovery, and translational medicine. It is designed to optimize scientific workflows with improved tools that provide greater understanding of chemistry, protein biology, and genomics. The model, according to OpenAI, “delivers the best performance on tasks that require reasoning over molecules, proteins, genes, pathways, and disease-relevant biology, and it is more effective at using scientific tools and databases in multi-step workflows such as literature review, sequence-to-function interpretation, experimental planning, and data analysis.”⁹

These conclusions were made after assessing GPT Rosalind’s ability to consider chemical reaction mechanisms; protein structures, mutation effects, and interactions; and phylogenetic interpretation of DNA sequences; to interpret experimental results, identify expert-relevant patterns, and synthesize external information to design follow-up experiments; and select and use the right computational tools, databases, and domain-specific capabilities to augment its reasoning.⁹ GPT-Rosalind outperformed other models using various public benchmarks.

“By supporting evidence synthesis, hypothesis generation, experimental planning, and other multi-step research tasks, this model is designed to help researchers accelerate the early stages of discovery,” according to the company. The goal is to “help life sciences organizations discover breakthroughs that wouldn’t otherwise be possible, with a much higher rate of success.”⁹

Named after Rosalind Franklin, who helped discover the structure of DNA, GPT Rosalind is intended to accelerate drug discovery by reducing the time it takes scientists to complete difficult, time-intensive work, not replace human researchers and the expert judgment they possess.¹⁰ Obtained results must, in addition, be confirmed through real-world validation.

GPT Rosalind is available in ChatGPT, Codex, and the API for qualified customers through OpenAI Group’s trusted access program.⁹ Only users with controls around eligibility, access management, and organizational governance will be considered to prevent potential biological misuse. A freely accessible Life Sciences Research Plugin for Codex is also available that helps scientists connect models to over 50 scientific tools and data sources.

Myriad benefits of AI have already been demonstrated for individuals, businesses of all sizes, educational/research institutions, nonprofit organizations, and governments. Like fire, the printing press, electricity, the Internet, and other “general purpose technologies,” AI is paradigm-shifting and will impact society on numerous levels, leading to major changes — at an unprecedented speed. Managing the risks posed by such rapid technological and societal change is essential.¹¹

In addition to grants made by the Life Sciences & Curing Diseases group, the OpenAI Foundation has been moving forward in the other key areas it is seeking to support. With respect to AI Resilience, efforts focus on addressing the new challenges that inevitably arise from more capable AI,” including making sure AI tools are “safe for young people and support health development,” helping strengthen the ability of society to prepare and respond to potential naturally occurring and AI-enabled biological threats, and ensuring the development of safer AI systems through effective testing, evaluation, standards development, and foundation research.3 The four specific initial focus areas outlined by the organization include bio-resilience, cyber-resilience, AI model safety, and AI’s impact on young people.¹¹

With regard to the impact AI may have on jobs and global and local economies, the OpenAI Foundation is “engaging with experts and communities — civil society, small business owners, unions, leading economists, policymakers, and others — to develop and fund practical solutions.”³ In June 2026, the Foundation announced its Jobs and Economic Futures program and indicated it was working to finalize over $130 million in grants through the AI Resilience program.¹¹

Two months later, the OpenAI Foundation announced the launch of its AI for Civil Society and Philanthropy program “dedicated to putting advanced AI to work on current challenges facing people and communities around the world.”¹² Its People-First AI Fund provides needed money to “trusted organizations” working in communities with high need to help them engage effectively with AI to improve their operations and impact. The goal is to provide sustained investment at scale, technical expertise, and support for responsible implementation through shared decision-making. The first $50 million in grants was awarded in 2025. A second round of funding, also $50 million going to 163 mostly new nonprofit awardees “exploring how AI can help communities across the United States access essential services, strengthen creative and cultural institutions, and support local journalism,” was announced in early October 2026.¹³ The unrestricted grants give “trusted institutions the resources and flexibility to explore how AI can strengthen their work — and how the people and communities they serve can engage with and shape its use.”

Separately, the OpenAI Foundation announced a $100 million investment in Common Health Coalition, a national multi-sector health collaborative, for creation of the Breakthroughs to Follow-Through (B2F), which will “work alongside state health departments, health systems, healthcare providers, and community organizations.”¹² First efforts will focus on doubling hepatitis C cure rates in Alabama, Illinois, Louisiana, and Massachusetts, using AI-enabled tools to help “identify and reconnect patients who have fallen out of care, synthesize records, track progress, and direct limited staff capacity where it can have the greatest impact” and make it possible for care teams to “reach more patients while reducing administrative burden.” The approach will eventually be broadened to other areas “where proven treatments and interventions are not reaching the people who need them, including HIV prevention and certain curable cancers.”

In addition to healthcare, the foundation is also working with organizations and communities to identify ways in which AI tools can help address problems in many different areas, such as agriculture, legal aid, and humanitarian response, with the people closest to these challenges defining the problems worth solving and leading and shaping how solutions are implemented, according to Anna Makanju, Head of AI for Civil Society and Philanthropy at The OpenAI Foundation.¹²

Finally, in September 2026, the OpenAI Foundation joined the Gates Foundation and partners around the world in a shared commitment⁠ to enhance AI model performance in low-resource languages, beginning with the development of data and infrastructure for voice AI where the gaps are largest and not likely to attract sufficient commercial investment.¹⁴

1. Taylor, Bret. “Built to Benefit Everyone.” OpenAI. 28 Oct. 2025.

2. “About.” OpenAI. Accessed 14 Oct. 2026.

3. Taylor, Bret. “Update on the OpenAI Foundation: A Note from Bret Taylor, Chair of the Board of Directors of the OpenAI Foundation.” OpenAI. 24 Mar. 2026.

4. “Reigniting the Discovery Engine for Tomorrow’s Cures: Why AI Matters for the Life Sciences.” OpenAI. Apr. 2026.

5. Callaway, Ewen. “How to Spend $25 Billion on Science: OpenAI’s Staggeringly Rich Research Charity Gets Under Way.” Nature. 658: 304–305 (2026).

6. Trefethen, Jacob. “AI for Alzheimer’s.” OpenAI Foundation. 8 Apr. 2026.

7. Mahajan, Abhishaike, and Jacob Trefethen. “Public Data for Health.” OpenAI Foundation. 15 Sep. 2026.

8. “Accelerating Life Sciences Research.” OpenAI Group. 22 Aug. 2025.

9. “Introducing GPT-Rosalind for Life Sciences Research.” OpenAI Group. 16 Apr. 2026.

10. Morrone, Megan. “OpenAI Launches New AI Model for Life Sciences Research.” Axios. 16 Apr. 2026.

11. Zaremba, Wojciech. “Resilience in the Age of AI.” OpenAI Foundation. 1 Jun. 2026.

12. Makanju, Anna. “AI for Civil Society and Philanthropy.” OpenAI Foundation. 13 Aug. 2026.

13. “Supporting Communities: Meet the 2026 People-First AI Fund Grantees.” OpenAI Foundation. 1 Oct. 2026.

14. Makanju, Anna. “Broadening the Benefits of AI, Starting with Voice.” OpenAI Foundation. 21 Sep. 2026.