The AI Oversight Gap, Siloed Systems, and What a Truly Inspection-Ready Company Looks Like: Q&A with Joseph Morwald

In July of this year, Krieger Scientific issued a release addressing the increase in 303 drug warning letters issued by FDA. In 2025, as new leadership took over the agency, FDA sent 303 drug warning letters, up 59% from the previous year.¹

Krieger’s founder and CEO Joseph Morwald believes that the increase is the result of structural problems within the industry. In his original statement, he said, “The FDA recommended outside compliance expertise in 87 percent of the warning letters it reviewed last year. That is not a coincidence. That is a diagnosis. When validation, calibration, and quality functions are managed in silos, by phase, by vendor, by department, with no single point of accountability, organizations lose the ability to catch problems before they become violations.”

Morwald spoke with Pharmaceutical Executive about the increase in warning letters and what he sees as the exact issues causing it. According to him, new technologies may be pushing companies away from human expertise, resulting in an increase in issues.

Click here for the full interview!

Pharmaceutical Executive: What is driving the increase of drug warning letters from FDA?
Joseph Morwald: The drivers behind what we're seeing are really a mixed bag. The industry is evolving very rapidly — the addition of AI tools, the shift in technology, the acceleration in drug approvals, and the change in experience levels at different levels within organizations all contribute to the factors driving today's compliance challenges.

In my opinion, one of the biggest drivers is that people are leaning more and more on AI and less on actual experience. It's easier to search for an answer than to actually know one. As a result, organizations are not utilizing the expertise of the generations before them, or even the experienced talent within their own organizations. AI is making people look smarter — but the reality is they're becoming less capable. We're relying less on actual intelligence and more on artificial intelligence, and that's showing up in warning letters and investigations that were produced without meaningful human review.

The FDA has identified this directly — it has made clear that there must be a human in the loop when reviewing, writing, creating, generating, and producing product. And that product can mean documentation, drugs, or vaccines. There is simply too much AI-generated material being produced without adequate human oversight, and we're increasingly relying on less experienced talent to do work that requires more experience.

PE: What structural problems are causing the industry's compliance failures?
Morwald: When we look at structural problems within companies, there's a lot of disjointedness and disorganization — siloed segments of the business operating without uniform systems and processes. There are too many computerized systems that don't talk to each other, with no APIs linking software together. The result is multiple databases tracking and managing similar data in ways that are fundamentally out of sync. I like to use the phrase: a man with two watches never really knows what time it is. The same problem occurs when you have multiple systems trying to track and manage the same information — if you have two databases, which one is correct? If you have three, you're really struggling.

Take asset management as an example. A single piece of equipment might be managed across three entirely separate systems — one from a finance perspective, one from a quality perspective, and one from a maintenance and calibration perspective. Each system is looking at the same asset but through a different lens. Finance is focused on depreciation schedules. Quality is focused on validation status. Maintenance and calibration are tracking whether the equipment is functioning and in spec. Some of the underlying data is the same — manufacturer, model, serial number, unique identifier — but it's being managed, run, and controlled by different groups, departments, and people. So those systems become disjointed and out of sync, and that is a major issue.

Historically, this was all managed on paper, and it was disjointed then too. But now we have less experienced people managing these silos, and the problem is compounding. Equipment and assets are not treated consistently across departments. Everyone has a singular function, and people are so focused on getting their own tasks completed that they're not thinking about the downstream impact on the rest of the organization.

Maintenance and calibration see that something is broken and fix it — without stopping to consider what the finance department is looking at from the same asset. Is it worth fixing? Is it past its depreciation schedule? Should it be replaced? When no one is looking at the full picture, spend becomes uncontrolled, replacements become reactive, and costs spiral. The organization is only ever seeing small pieces of a problem that needs to be viewed as a whole.

PE: What is the significance of FDA sending its first warning letter about inappropriate use of AI?
Morwald: When it comes to AI, there is a significant amount of phantom and ghost information in the space — and when something is generated by an AI tool, you must verify what is real and what is not. We are seeing significant examples of references in documentation that are completely fabricated. This problem exists beyond life sciences, but in life sciences it carries a different weight entirely. We are talking about people's lives — health and safety, life and death — in the manufacture of pharmaceuticals, vaccines, and medical devices. If we get it wrong, people die.

If we are not verifying reference material, not conducting the right tests, not producing accurate and factual data — if we are relying on made-up information, information that doesn't apply, or missing critical steps entirely — the consequences are fatal. That is not hypothetical. That is the reality of this industry.

The FDA recognizes this, and their position is clear: there must be a human in the loop to close the circle on anything generated by AI. It has to be reviewed. It has to be verified. And it must be credible.

PE: How would you describe a company that is well-prepared for regulatory inspection?
Morwald: Companies that are well-positioned for the future are the ones at the forefront of regulatory compliance — continuously improving, actively engaging with the regulatory framework, and understanding that the FDA is a partner in the delivery of safe products, not an adversary.

One of the most important things the FDA has consistently emphasized is the value of unbiased, third-party independence. A company should not be doing everything internally. You need an outside perspective to remove bias, because it's too easy for internal teams to look at their own work and conclude that it's fine. The question is never simply whether the internal team understands the data, or whether the FDA understands it. The standard is whether any informed person — a common reviewer with no stake in the outcome — can look at the data and confirm that it is what it claims to be, and that it is good.

A company that is truly well-positioned, both today and for the future, will use all available tools — including AI — but will also validate and verify the results those tools produce. It will continuously improve, and it will keep raising the bar on regulatory compliance rather than treating it as a fixed target to meet and move on from.

Source

1. FDA Warning Letters Surge 59 Percent. One Life Sciences Firm Says the Industry Is Solving the Wrong Problem. Krieger Scientific. July 9, 2026. Accessed August 5, 2026. https://www.einpresswire.com/article/925352156/fda-warning-letters-surge-59-percent-one-life-sciences-firm-says-the-industry-is-solving-the-wrong-problem