Once FDA read with AI, sponsors’ inconsistencies got cheaper to find
FDA’s Elsa gave reviewers a fast tool for searching and comparing submissions. Sponsors needed three disciplines in response: dossiers a machine reader cannot trip over, an evidence trail that corrects misreadings quickly, and validation records for their own AI.
Looking Back revisits one week at a time, from January 2020 to today, and asks what it meant for the banks, health systems and public agencies we serve. The series was written in 2026, with the benefit of hindsight, and each article says which week it looks back on.
The week
On June 2, 2025, the Food and Drug Administration announced the agency-wide launch of Elsa, a generative AI tool meant to help employees, from scientific reviewers to investigators, work more efficiently. FDA said Elsa could help read, write and summarize documents, compare labels, summarize adverse events and generate code for internal databases, and could help identify high-priority inspection targets. The agency said the tool ran in a secure government cloud environment and did not train on data submitted by regulated industry. It said the launch came ahead of schedule, after a generative AI pilot in scientific review. In May, Commissioner Marty Makary had set a goal of scaling AI across all FDA centers by the end of June.
Life sciences companies have spent years asking how they should use AI. Elsa reversed the direction of the question. From June 2025, the people reviewing a submission had a generative AI assistant at hand. FDA did not say Elsa made decisions, and sponsors cannot see how any reviewer used it. They can still reason about what changes when the reader has a fast tool for summarizing, comparing and searching. We see three consequences, each with a different owner. They are what a machine reader surfaces in the dossier and how a sponsor corrects a machine misreading. The third is how the sponsor’s own AI use will be judged by a regulator that now uses the same technology.
What a machine reader surfaces in the dossier
A marketing application in the Common Technical Document format can run to many thousands of pages across five modules. A human reviewer samples it. A reviewer with a summarization tool can ask for every mention of a dose, an endpoint or an exclusion criterion and see them side by side in seconds. That changes the economics of finding inconsistency. A safety population defined one way in the clinical summary and another way in a study report used to take a careful reader to notice. Now it can surface from a single prompt.
The practical response sits in regulatory operations. Sponsors can keep a controlled vocabulary for each program: one definition per endpoint, population and analysis set, used in every module. They can run their own automated cross-module checks before filing, comparing numbers in summaries against source tables. They can treat the proposed label, the integrated summary of safety and the clinical overview as one linked set, with a change log showing when a figure was updated everywhere. None of this is new advice. What has changed is the cost of skipping it, because the reviewer’s cost of catching it has fallen.
There is a trade-off. Rigid consistency checks slow late-stage document work, and submissions are often assembled under deadline. A sensible compromise is to aim automated checks at the facts most likely to be compared: doses, sample sizes, adverse event counts and endpoint definitions. Stylistic variation can wait.
How a sponsor corrects a machine misreading
Generative tools produce confident errors. A summary can merge two studies, invert a finding or cite a passage that does not exist. FDA described Elsa as an assistant to staff, and reviewers remain responsible for their conclusions. Even so, a sponsor may receive an information request that appears to rest on a misreading, and it will rarely be told whether a tool was involved.
The defense is an evidence trail that makes correction fast. That means a submission index mapping every key claim to its location by module, section and table, maintained by the regulatory team through the whole review cycle. It means response templates that quote the source passage instead of paraphrasing it. And it means answering questions with citations a reviewer can verify in one step. A sponsor that can say the figure appears in a named table of Module 2.7.4 and matches the study report resolves a misunderstanding in one round. A sponsor that has to reconstruct where a number came from loses days, and review clocks do not stop for that.
Sponsors should also avoid the opposite error of assuming every odd question came from a tool. Speculating about AI in correspondence with a regulator helps nobody and can sour a working relationship. The goal is a record that works whoever, or whatever, did the reading.
How the sponsor’s own AI use will be judged
In January 2025, FDA issued draft guidance on using AI to support regulatory decision-making for drugs and biological products. It proposed a risk-based credibility assessment: define the question the model addresses, define its context of use, assess model risk, then plan, carry out and document the work that establishes credibility for that context. The guidance targets AI that produces information supporting decisions about safety, effectiveness or quality. It does not cover AI used only for operational efficiency, such as drafting assistance.
That boundary matters, and it blurs easily in practice. A tool that drafts narrative text is an efficiency aid. A tool that codes adverse events, selects patients from real-world data or generates tables for the integrated safety summary is producing evidence. Sponsors need an inventory that places each AI use on one side of that line, with an owner in quality assurance who signs off the classification and revisits it when the use changes.
For uses on the evidence side, the record should show the model version, the data it was tested on, the error rates found and the human review applied. A regulator whose staff now work with a generative tool, and have seen its errors firsthand, is well placed to ask precise questions about how a sponsor caught its own tool’s mistakes. Sponsors that answer with a validation record, instead of a policy statement, will spend less review time on process questions. The cost is real: validation takes data science time that would otherwise go to new uses. That cost is lower before filing than during a review.
The three parts call on different functions: regulatory operations for the dossier, the regulatory affairs lead for the evidence trail, and quality with data science for the AI inventory. What links them is that each produces records a skeptical reader can check quickly. That was always good practice. From mid-2025 it was also the fastest route through a review with software in the loop.
With hindsight
The tool drew scrutiny within weeks. In July 2025, CNN reported that several current and former FDA staff said Elsa could invent studies or misrepresent research. Agency leaders acknowledged that it could hallucinate and said its use was optional, while other reporting described it as useful for routine tasks such as summarizing meetings. That mix supports the defensive part of the argument above: sponsors needed records that make a misreading easy to correct.
The policy direction on sponsors’ own AI use also held. In January 2026, FDA and the European Medicines Agency published ten shared principles for good AI practice across the medicines lifecycle, built around risk, context of use and documentation. By this point in 2026, the expectation that sponsors can show how AI-generated evidence was validated is shared by the two largest drug regulators. Companies that classified their AI uses early and kept their dossiers internally consistent are better placed for the questions those principles imply.