Artificial intelligence is rapidly becoming part of everyday quality and regulatory activities. It can help draft procedures and specifications, generate risk management documentation, and create quality records. The numerous AI tools available can significantly improve efficiency.
However, a recent FDA warning letter reminded us that while AI can support compliance activities, it cannot replace human expertise, oversight, or accountability.
In an April 2026 warning letter issued to Purolea Cosmetics Lab [Purolea Cosmetics Lab - 722591 - 04/02/2026 | FDA], the FDA cited the company's inappropriate reliance on AI-generated content to create specifications, procedures, and manufacturing records intended to meet regulatory requirements. Investigators found that the company failed to review AI-generated documents adequately and had not identified critical regulatory requirements, including process validation.
When questioned, the company stated it was unaware of these requirements because its AI tool had not identified them. The FDA has made this clear through this warning letter: this is not an acceptable justification for noncompliance.
While the warning was directed at a pharmaceutical manufacturer, the message is equally relevant to the medical device industry. AI tools can generate convincing and professional-looking quality system documents, but they do not understand regulatory intent, the company’s product-specific risks, or the unique requirements of an organisation's Quality Management System (QMS).
AI outputs may be incomplete, inaccurate, outdated, or fail to capture critical regulatory requirements. If these deficiencies are incorporated directly into procedures, technical documentation, validation protocols, or risk management files, they can create significant compliance gaps that may not be discovered until an audit, inspection, or adverse event occurs.
For pharmaceutical and medical device manufacturers, the principle remains unchanged: responsibility for compliance cannot be delegated to software. Any AI-generated content must undergo the same level of review, approval, and control as information created by human personnel. Qualified individuals must verify that documents are accurate, complete, and appropriate for the intended use before approval and implementation.
The real risk is not AI itself, but overreliance on AI without effective controls. Because it has proven to be an effective and efficient tool when used correctly, organisations should establish clear governance for AI use within their QMS. This includes defining acceptable use cases, ensuring qualified personnel review all AI outputs, maintaining document control, validating AI-assisted processes where appropriate, and training employees on the limitations of AI-generated content.
The FDA's warning highlights an important reality for the life sciences industry: regulators expect organisations to leverage technology responsibly, not blindly. AI can be a powerful productivity tool, but it is not a substitute for regulatory knowledge, professional judgement, or quality oversight. As AI adoption accelerates across organisations, those that implement strong review processes and maintain clear human accountability will be best positioned to realise AI's benefits while avoiding potentially serious compliance consequences.
Establishing a robust review process is, however, easier said than done with the plethora of AI tools and solutions now available. An effective way to tackle this issue is to conduct a comprehensive risk assessment for each AI use case within your organisation, then focus your efforts on the highest-risk ones.
To support this assessment, SeerPharma has formed a strategic partnership with AIQURIS.
AIQURIS provides a structured way for organisations to examine an AI use case before and during deployment. It considers risks across areas including safety, security, ethics, legal requirements, performance, and sustainability, and helps you identify potential concerns and the controls needed to manage them.
You need to start with a policy, identify what AI is already in use, and make sure that all your personnel are trained in responsible use of AI.