The Food and Drug Administration has gone all in on AI since Martin Makary, M.D., took over the agency, most recently demonstrated by the coronation of new generative AI tool Elsa. Now, Makary and Vinay Prasad, M.D., director of the FDA’s Center for Biologics Evaluation and Research, have published an outline of the FDA’s priorities that includes speedier drug reviews powered by AI document evaluation.
The duo believes that AI can make a “first-pass” at reviewing the documents included in applications received by the agency, which can sometimes reach more than 500,000 pages per application, Makary and Prasad wrote in the plan, as published June 10 in the Journal of the American Medical Association.
In addition to using generative AI tools, Makary and Prasad also plan to speed up approval review times by asking applicants to submit materials like manufacturing plans, draft labels and packaging before clinical trials for the proposed new product are complete.
“The time from when pivotal trial results are known to when decisions are made must be shortened,” the authors wrote. “This has implications for public welfare and will improve the risk-reward calculation of drug development.”
"At the FDA, we aim to pilot programs to deliver sponsors a decision in weeks," the pair added.
Makary and Prasad pointed to the COVID-19 pandemic, when the FDA put forth a speedier review process for therapies to treat the new virus, as a "clear demonstration that rapid or instant reviews are possible." The COVID-era program did not mention the use of AI to speed up reviews.
AI can also help reduce the need for animal testing by powering computational models of drug toxicity, the authors wrote, referring back to an agency priority announced earlier. In April, the agency shared plans to shift away from animal testing requirements for new monoclonal antibodies.
The FDA’s AI tool Elsa, which was recently released a month ahead of schedule, is already being used in clinical protocol reviews, scientific evaluations and to identify high-priority inspection targets, according to the agency. Elsa is meant to be adopted throughout the FDA’s entire organization.
Elsa’s early release may have been premature, as reports have rolled in of the tool being buggy and outdated.
Asked to name FDA leaders, Elsa returned answers of staff who are no longer at the agency, according to a report from Stat. And when asked to summarize information about FDA-approved products, Elsa gave responses that were only partially accurate at best, an insider told NBC News.
It’s also unclear how much Elsa will help the FDA reduce review times. Recent layoffs at the agency have included staff that work on drug reviews, and since January, the agency has already missed several PDUFA dates, including for GSK’s Nucala in chronic obstructive pulmonary disease, Novavax’s protein-based COVID-19 vaccine and Stealth BioTherapeutics’ treatment for an ultra-rare disease.
Elsa’s launch also came on the heels of Trump’s 2026 budget proposal (PDF), which recommends cutting the FDA’s budget by 4%, including the cutting of 1,940 staff salaries.