FDA bets on AI and cloud monitoring for clinical trials as it looks to rebuild after DOGE layoffs

The FDA is launching a pilot program to monitor clinical trials in real time using AI and cloud computing, a move the agency says could dramatically shorten the time it takes to approve new drugs.

The U.S. Food and Drug Administration has announced its first real-time monitoring program for clinical trials. Instead of waiting for pharmaceutical companies to submit documentation that can run millions of pages long, the agency will receive direct data feeds from ongoing studies through AI and cloud infrastructure.

FDA Commissioner Marty Makary called the announcement a milestone. Right now, it takes ten to twelve years to bring a new drug to market, and roughly 45 percent of the time between the first clinical phase and regulatory submission is spent on paperwork and administrative tasks. The new pilot program gives the FDA a direct data feed from active trials, so when a patient develops a fever or a tumor shrinks, regulators can see it happening in the cloud in real time, Makary said.

Jeremy Walsh, the FDA's Chief Artificial Intelligence Officer, estimates the approach could cut 20 to 40 percent off total trial duration. Walsh stressed that safety standards would not be lowered. The idea came up last summer, he said, since the review process has barely changed since the 1960s. AstraZeneca and Amgen are the first partners in the pilot. The FDA has also published a public request for information to gather additional proposals for AI-driven improvements to clinical trials.

Over 80 percent of FDA staff now use generative AI

The pilot also reflects the reality of deep cuts imposed by the Trump administration and DOGE. All consolidation efforts are being carried out without additional staff or resources, following significant layoffs in early 2025. Makary said the annual savings of at least $120 million would fund, among other things, the rehiring of up to 3,000 scientists.

The agency has also seen a dramatic jump in AI adoption. In early 2025, only about one percent of FDA staff regularly used generative AI, according to Walsh. Today, that number is above 80 percent. An internal tool called Elsa helps employees read, write, and summarize reports. In early pilot projects, AI reduced administrative tasks that used to take ten days down to 20 minutes.

However, several FDA employees told CNN last summer that Elsa regularly fabricates nonexistent studies and misrepresents research data. Walsh acknowledged at the time that Elsa is no different from many other large language models and generative AI tools, and that it can hallucinate.