OPENAI has held a series of discussions with US Food and Drug Administration (FDA) officials on the potential use of artificial intelligence to rationalize parts of the drug assessment process, according to a WIRED report published on Wednesday.
The talks mark a significant step towards the integration of AI generative tools into one of the most closely regulated and scientifically rigorous fields of the American federal government.
At the center of the conversation is a project called CDERGPT, which seems to be an experimental AI system aimed at helping the FDA (CDER) medication assessment and research center. The CDER is the main division of the agency to supervise the safety and efficiency of prescription and over-the-counter medication in the United States.
Register For TEKEDIA Mini-MBA Edition 17 (June 9 – September 6, 2025)) Today for early reductions. An annual for access to Blurara.com.
Tekedia Ai in Masterclass Business open registration.
Join Tekedia Capital Syndicate and co-INivest in large world startups.
Register become a better CEO or director with CEO program and director of Tekedia.
According to Wired, meetings involved not only OPENAI staff, but also the participants of a newly formed federal agency called the Ministry of Government Effectiveness, or DOGE, a broader initiative of Elon Musk to reform the functioning of government systems thanks to technological innovation. The exact role of Doge in these conversations remains uncertain, but sources familiar with the talks confirmed its presence during several meetings with the FDA.
The role of AI in the acceleration of regulatory sciences
The FDA would have studied how artificial intelligence could accelerate certain parts of the drug development calendar, especially during the last stages of the journal, where the volumes of clinical data and scientific documentation must be processed. In the United States, the development of drugs can extend to more than a decade, from preclinical stages to the approval of the FDA, the regulatory examination often representing one of the most prolonged phases.
AI could help reduce this window. The FDA finished its first examination helped in AI-AI of a scientific submission for a product. The examination was characterized as an important step in the agency’s modernization efforts.
“The FDA is committed to supporting innovative approaches to the development of medical products by providing an agile and risk -based framework that promotes innovation and guarantees that the agency’s robust scientific and regulatory standards are respected,” said FDA commissioner Robert Califf. “With the appropriate guarantees in place, artificial intelligence has a potential transformer to advance clinical research and accelerate the development of medical products to improve patient care.”
Jeremy Walsh, a former AI specialist from the Ministry of Defense, recently appointed as an officer of the FDA AI, directed the agency’s dialogue with Openai. Walsh would also have developed internal advice on how the FDA could adopt models of large languages (LLM) such as GPT-4 for the regulatory sciences.
Although the CDERGPT project remains largely experimental, people close to talks have told Wired that conversations focused on how such a system could help in the review of the literature and perhaps even to write parts of regulatory documents – tasks that are often repetitive and long.
The promise and traps of the AI generating in the Revue des drugs
Despite the enthusiasm around AI, many experts warn against risks. The reliability of large language models, especially in the fields with high issues like medicine, is always under control. Generative AI models like those of Openai have demonstrated incredible capacities, but they also have real limits with regard to hallucinations, transparency and reproducibility.
This backdrop has created a real need for political advice on appropriate data sets for training models used in this type of work, how performance should be measured and if the results are interpreted by human examiners.
According to the wired report, the FDA actively invests in understanding these challenges. The agency has launched a scholarship program to train experts on the use of AI in regulatory sciences and has experienced its own models in large internal language to explore specific applications in areas such as precision medicine and post-commandcal surveillance.
A transition to the level of the government to AI?
The involvement of DOGE, a government body would have been initiated under the influence of Elon Musk, adds another dimension to history. Although DOGE’s official mandate is still largely indefinite, it has been described by internal sources as an inter-agencies working group aimed at making government more effective through emerging technologies.
DOGE’s participation in FDA talks with OPENAI suggests an increasing federal appetite for the transformation led by AI, not only in public services, but within scientific agencies that have traditionally worked with high levels of prudence and bureaucratic inertia.
Time will tell us if the OPENAI CDERGPT or similar projects become central devices from the FDA regulatory toolbox. But the will of the two parties to explore even such a future underlines a significant change in the way in which the American government thinks of innovation, automation and public health.
