Another Food and Drug Law Institute (FDLI) Annual Conference is in the books, and the Food and Drug Administration’s (FDA) messaging at the 2026 edition blended big-picture aspirations with clear near-term change. Right from the outset, FDA Commissioner Dr. Marty Makary announced a new inspection pilot that uses AI to identify “low-risk” facilities for “one-day inspections.” He also flagged near-term priorities, largely aspirational, aimed at speeding development and review, modernizing clinical trials and the investigational new drug (IND) process (including changes to the institutional review board (IRB) model), expanding patient access to experimental treatments (including psychedelics), and streamlining FDA’s internal decision-making with technology. Overall, agency leaders projected confidence—downplaying staffing concerns, emphasizing productivity, and expressing optimism that innovation will translate into better regulatory outcomes. Notably absent during the opening sessions was any meaningful discussion of direct-to-consumer (DTC) advertising by the agency, despite the attention it has received from this administration.
The headline: AI-assisted “one-day inspections”
Dr. Makary’s most concrete announcement was the launch of a one-day inspectional assessment pilot program. FDA will use AI-enabled risk scoring to identify lower-risk facilities and conduct shorter, targeted assessments. To be sure, this is not a wholesale replacement of traditional inspections or the need for the same, as clarified after the initial remark by Dr. Makary. Rather, it appears to be a change in deployment. FDA is testing whether streamlined, risk-informed assessments can screen efficiently and reserve more intensive inspections for higher-risk situations. Facility selection will reflect risk factors such as product type, inspection history, and operational profile. On social media, Dr. Makary said that these have already begun and have been largely successful.
The most straightforward takeaway is that, given the reported shortage of reviewers, FDA may need to prioritize inspections of higher-risk facilities. So, to make the best use of limited resources, a one-day inspection of a low-risk facility is less burdensome. However, the bigger question—raised by us and others at the conference through informal discussion—is how AI will identify “low-risk” facilities. What criteria will it use? Will the designation be based on prior inspection history, known activities, or some combination of factors? And, as with many AI-driven approaches today, can this method be trusted? All this, and more, to come.
Overhauling INDs
Dr. Makary was blunt about the IND timeline gap: He cited successes in process streamlining in Australia and a projected 60-day timeline in China for new drug approval versus more than 300 days in the U.S. His stated plan is an overhaul of the IND process using a mix of rulemaking, guidance, and clearer communications with sponsors, noting that this is a significant lag in the process.
Dr. Makary focused specifically on IRBs, expressing a sense of frustration over how long they can take and even suggested that sponsors should independently find their own review boards. As Dr. Makary acknowledged, his perspective appears shaped in part by anecdotal accounts of IRBs stagnating the research process and by his own experience as a researcher.
However, in the US, FDA regulations require IRB approval or oversight in certain areas, but the IRB itself is not traditionally governed by the same regulations that govern the sponsors whose study plans are reviewed and approved by IRBs. Instead, IRBs generally operate to uphold paradigms of ethical practices, implementing global standards for human subject research, rather than to govern review timelines or efficiency incentives. So, barring removal of IRB requirements from the regulations, broadly granting waivers, or putting into place some other incentivization system for IRBs to work faster, it is not entirely clear how this initiative will be effectuated.
Modernizing clinical trials—“real-time” oversight
Dr. Makary also previewed a real-time clinical trial pilot enabled by cloud computing. The concept is that adverse events can be reported and reviewed as they occur, rather than arriving in periodic batches. He framed this as an attack on “dead time” in trials—time lost to lagging data flow. If FDA can evaluate safety signals in near real-time—and ensure negative results are reported—trials can be monitored more efficiently, with near real-time decisions about trial status.
Like the IND streamlining process, it remains to be seen whether and to what extent this will move things forward. The industry is interested, though, as Dr. Makary cited a few household names taking part—large companies that have already opted into this program. At this point, FDA’s real-time review will be limited to safety and efficacy endpoints that are agreed upon with sponsors in advance. But further questions remain about how real-time monitoring will work in practice, whether and to what extent blinding will be impacted, how the monitoring will interface with data monitoring and safety committees that may already be in place, and much more.
Faster decisions on experimental therapies—including psychedelics
Dr. Makary’s comments on psychedelics were among his most candid. He did not claim certainty that they are effective for psychiatric conditions—particularly in veterans—but he stressed the need to reach decisions quickly and argued that patients deserve the chance to try them. He also pointed to the scale of unmet need and the volume of patient testimony, suggesting FDA should not dismiss those factors simply because the mechanisms are not fully understood—an approach that departs from FDA’s typically narrow and rigid posture toward experimental treatments.
This is a talk track that has been repeated a few times over the past few weeks, but it bears mentioning because it signals a meaningful shift in FDA's stance toward patient testimony and unmet need as factors in regulatory decision-making—one that could have broader implications for experimental care beyond psychedelics.
IT and AI to streamline review and reduce friction
A consistent throughline was “common sense” modernization—especially around FDA’s data infrastructure. Speakers from the agency highlighted consolidation of adverse event reporting as a practical win: better data quality, less FOIA burden, and more staff time redirected to science.
Speakers also reinforced that AI tools are speeding literature review, helping reviewers find relevant precedent, and strengthening both pre- and post-market decision-making.
As agency speakers noted, tasks like sorting data and searching prior approvals have long been time-consuming. So while these modernization efforts may be less headline-grabbing, they could meaningfully improve efficiency and make prior precedent more accessible, thus more impactful in current dialogue.
What didn’t make it: DTC advertising
Given the administration’s enforcement focus on DTC advertising, its near-total absence from the agency’s conference messaging stood out. Whether that reflects strategic silence, internal prioritization, or timing, it was difficult to square with recent enforcement attention. Later panelists were keen to point out this notable absence.
We’ve been monitoring this space closely and have covered recent surges in enforcement around DTC drug advertising, including in early 2026.[1] Many observers expected this administration to significantly curb—or even attempt to effectively eliminate—DTC drug advertising, and it has been widely viewed as an agency priority to police promotional practices, particularly around safety presentation. Many also anticipated heightened enforcement focused on telehealth. While enforcement trends suggest otherwise, the agency silence may temper speculation that DTC advertising is a top priority going forward.
Final Thoughts
FDLI 2026 showed an FDA trying to do two things at once: speed up its work and modernize—without clearly acknowledging whether limited staffing and resources are part of what’s driving these changes. The key questions in the near term are straightforward: Can AI reliably flag “low-risk” facilities for one-day inspections? What definitional changes to the IND and IRB processes will actually shorten timelines? And can “real-time” trial monitoring cut delays without weakening oversight? What FDA did not highlight—especially DTC advertising—may be just as telling as what it did.
FOOTNOTES