FDA Explores AI Collaboration with OpenAI to Accelerate Drug Approval Process

This development appears to be part of a broader FDA initiative to modernize and expedite the drug approval timeline using emerging technologies.

The U.S. Food and Drug Administration (FDA) is exploring deeper integration of artificial intelligence (AI) into its regulatory review process, with recent meetings held between agency officials and representatives from OpenAI, the developer behind ChatGPT, according to sources familiar with the matter.

This development appears to be part of a broader FDA initiative to modernize and expedite the drug approval timeline using emerging technologies.

Makary Hints at AI-Driven Regulatory Reform

FDA Commissioner Marty Makary signaled the agency's evolving stance on AI in a recent post on social media platform X, asking:

“Why does it take over 10 years for a new drug to come to market? Why are we not modernized with AI and other things?”

Makary noted that the agency had just completed its first AI-assisted scientific review for a product, describing it as “just the beginning.” His comments followed his remarks at the American Hospital Association’s annual meeting, where he spoke about AI’s potential role in accelerating the approval of treatments for diabetes and certain cancers.

While Makary did not mention OpenAI by name, insiders confirmed that a small team from the company has met with FDA officials in recent weeks to discuss a project reportedly titled “cderGPT”—a likely reference to the FDA’s Center for Drug Evaluation and Research (CDER)—as well as a related initiative known as “Research GPT.”

OpenAI, Elon Musk Associates, & the FDA’s AI Officer

Sources said the meetings involved not only OpenAI personnel but also two individuals affiliated with Elon Musk’s so-called “Department of Government Efficiency.”

Leading the discussions from the FDA’s side is Jeremy Walsh, the agency’s first-ever Chief AI Officer, who was appointed to spearhead AI adoption across regulatory workflows.

Walsh has also held conversations with Peter Bowman-Davis, a Yale undergraduate currently on leave to serve as the acting Chief AI Officer at the Department of Health and Human Services (HHS).

Bowman-Davis is also affiliated with Andreessen Horowitz’s American Dynamism team, a group focused on tech-enabled governance and public-sector transformation.

No contracts have been finalized between OpenAI and the FDA as of now, and OpenAI declined to comment on the reported meetings.

A New Era of AI-Enabled Drug Review?

Former FDA Commissioner Robert Califf, who served from 2016 to 2017 and again from 2022 until January 2024, confirmed the agency has already been using AI in various review-related functions.

“It will be interesting to hear the details of which parts of the review were ‘AI-assisted’ and what that means,” Califf said in an email.

He added that while final drug approval reviews are a focal point, AI has broader potential across the entire regulatory pipeline. “Final reviews are only one part of a much larger opportunity,” Califf noted, pointing to AI’s utility in document analysis, post-market surveillance, and workflow automation.

Despite these advances, Califf acknowledged that AI alone cannot solve the complexities of the drug development lifecycle, noting that most drug candidates fail during clinical trials and never even reach the FDA review phase.

Industry Calls for Guardrails on AI Use

Experts in precision medicine and healthcare AI have cautiously welcomed the FDA’s interest in AI but emphasize the need for policy oversight and transparency.

Rafael Rosengarten, CEO of oncology-focused startup Genialis and board member of the Alliance for AI in Healthcare, said he supports automating routine tasks but urged that AI systems be trained with high-quality, domain-specific data and held to rigorous performance standards.

“These machines are incredibly adept at learning information, but they have to be trained in a way so they're learning what we want them to learn,” Rosengarten said. He sees immediate potential for AI in administrative tasks, such as checking whether regulatory applications are complete before a full review.

“Something as trivial as that could expedite the return of feedback to the submitters based on things that need to be addressed,” he explained. More advanced applications—like AI-guided efficacy assessments—would require extensive validation before use.

A former FDA staffer added a note of caution, expressing skepticism about using large language models (LLMs) like ChatGPT in clinical settings due to their tendency to generate convincingly incorrect or fabricated information. “Who knows how robust the platform will be for these reviewers’ tasks,” the ex-staffer said.

Existing FDA Fast-Track Mechanisms Still Central

The FDA already employs several accelerated pathways to approve promising new drugs. These include:

Fast Track designation, which expedites review for drugs that address unmet medical needs in serious conditions.

Breakthrough Therapy designation, which offers priority review for drugs showing substantial improvements over current treatments.

However, these fast-track options still require human-led evaluations, and AI integration could enhance rather than replace these processes. As FDA spokesperson Andrew Powaleny of industry group PhRMA noted, “Ensuring medicines can be reviewed for safety and effectiveness in a timely manner to address patient needs is critical.”

“While AI is still developing, harnessing it requires a thoughtful and risk-based approach with patients at the center,” he added.

FDA’s Broader AI Ambitions

The agency’s AI ambitions extend beyond the current pilot. In December 2023, the FDA announced a fellowship to support research into large language models for internal applications.

The selected fellow would explore LLM use cases in precision medicine, drug development, and regulatory science, according to the job description.

In January 2024, OpenAI launched ChatGPT Gov, a self-hosted version of its popular chatbot, tailored for U.S. government compliance.

The company also revealed it is pursuing FedRAMP Moderate and High authorizations for its ChatGPT Enterprise platform. These certifications are mandatory for handling sensitive federal data, and without them, an AI service cannot store or process government information.

Stay tuned for more such updates on Digital Health News.