This blog specifically relates to submissions to the MHRA Compliance Teams following GxP inspections.
The MHRA has identified examples of information relating to inspection activity submitted to compliance teams which have been generated by Artificial Intelligence (AI). This blog describes our current thinking in relation to the use of AI when generating information to support inspections.
The reality
AI tools are being used to draft inspection responses supplied to the MHRA. The technology offers genuine benefits: helping articulate complex technical issues, improving consistency, speeding up routine drafting and enabling innovation. Used properly, AI can support better regulatory outcomes and improve patient safety.
However, we have also encountered responses containing references to MHRA guidance that doesn't exist, citations of inappropriate regulatory frameworks, and responses to serious deficiencies that appear designed to mislead rather than address underlying problems. For example, one inspection response received by the MHRA totalled over 90 pages and did not address the deficiencies identified. This resulted in a disproportionate use of Inspector time and expertise to review and respond. In at least one case, an AI-generated response to a serious, patient safety impacting deficiency contained material inaccuracies, including non-existent references as well as inaccurate information that delayed the resolution of serious compliance failures, creating additional administrative burden and delays. The contents of this response required review from multidisciplinary teams, as well as a full analysis of previously issued MHRA guidance, resulting in the time needed to review and respond increasing from around 4 hours to over 20 hours. Aside from being a strain on resource, the use of AI being applied inappropriately has shifted from theoretical risk into actual, realised risk.
What hasn't changed
Organisations have always been responsible for the accuracy of their responses to the MHRA and persons responsible for them have always been expected to verify factual claims. Technical review remains a core component of quality systems. Similarly, materially false statements to inspectors have always been regulatory offences. These responsibilities do not change because drafting tools have.
Clarifications
The MHRA maintains its position as a regulator enabling innovation focused on safe and effective regulation and patient safety as our number one priority. To that end our concern isn't whether you use AI; it's whether your submissions are accurate, verifiable, and prepared under appropriate oversight.
We're issuing this blog post to make this explicit and focus on process and outcomes rather than specific tools. The principles are simple:
All submissions/responses must be:
- factually accurate and verifiable
- technically reviewed by appropriately experienced people
- signed off by someone with authority and accountability
- supported by evidence for factual claims
- appropriate to the specific regulatory context.
These expectations apply regardless of how you draft your submissions. Good quality management systems work whether you use AI, templates, consultants, or write from scratch.
Voluntary disclosure
In our evolution of understanding AI use, we are offering organisations the option to disclose AI use in responses/submissions to the compliance teams. This isn't mandatory, but we believe transparency benefits everyone.
If you choose to disclose:
- include a brief statement at the start of your submission/response to Compliance Teams
- identify which sections involved AI assistance
- confirm human verification and approval
- transparency in the use of AI tools, alongside robust mechanisms to ensure accuracy and reliability of submissions, indicates a more mature and transparent quality culture.
Companies using robust verification processes, whether they use AI or not, demonstrate the kind of quality thinking which protects and prioritises patients. Disclosure helps us understand how the sector is evolving and allows us to provide better support to industry. Inspectors will consider this positively when assessing organisational compliance.
What inadequate verification/quality check of submitted material looks like
When we review responses, we assess the quality of oversight processes. Warning signs of inadequate verification of the information submitted include:
- factually incorrect statements or non-existent references
- generic language inappropriate to specific circumstances
- lack of organisation-specific detail where expected
- citations of regulatory frameworks without explanation of relevance
- inconsistent technical terminology across submissions
- overly verbose responses failing to deal with the subject matter appropriately
These patterns, whether caused by AI or other factors, indicate quality system weaknesses. Where we identify inaccurate information that affects the conduct of an inspection then we may act; if the information supplied in response to an inspection is inaccurate, incomplete or overly verbose then we may reject the response or return it for another attempt. We may also consider the organisation higher risk in terms of future inspection prioritisation due to poor CAPA and or refer the organisation to the Inspection Action Group (IAG). We may also assess your verification processes and repeated patterns may indicate systemic failures warranting regulatory action.
Working together on responsible AI use
MHRA have discussed previously its interest in transparency within regulation. We are setting outcome expectations (accurate, well verified documentation) and allowing industry to determine methods, which has been consistent with our approach to MHRA regulation and industry interpretation. Companies using robust processes, AI-assisted or otherwise, meet regulatory standards. Those submitting unverified material don't.
As AI becomes more sophisticated, this framework remains relevant because it focuses on accountability and process, not detection and prohibition. We are not trying to police technology used in the inspection process; but we are ensuring a focus on patient safety and proportionate regulatory responses through appropriate quality oversight.
This approach is consistent with the MHRA's broader interest in outcome-focused, regulation, where we set clear expectations and work collaboratively with industry to achieve them.
This is about proportionate risk management
Responses to inspection findings should demonstrate effective corrective and preventive action: identifying genuine root causes through evidence-based investigation, assessing impact proportionately, implementing actions that address underlying issues rather than symptoms, and verifying their effectiveness. AI tools may support elements of this process but cannot substitute for the technical understanding and organisational knowledge required to develop meaningful corrective actions. Where responses demonstrate superficial or generic CAPA without evidence of genuine root cause analysis, this will be identified as a quality system weakness regardless of how the response was drafted.
MHRA have written previously on the subject of what effective CAPA looks like.
We recognise AI capability will advance rapidly, as it does the MHRA will continue to build a framework based on setting clear expectations, transparent processes and shared responsibility. Those who use AI responsibly will thrive, whereas those attempting to use technology to obscure inadequate responses, fill resource or expertise gaps, or hide a lack of knowledge causing serious deficiencies will be scrutinised.
Leave a comment