Use of AI in GXP Inspection Responses

Key changes

Establishes expectations for the use of AI in regulatory communications. Emphasizes that MAHs and sponsors are fully responsible for the accuracy of AI-generated content and warns against inaccuracies or 'hallucinations' in inspection responses.

PV impact

Impacts inspection readiness and CAPA workflows. Organizations using AI for drafting responses must implement quality checks to ensure citations and factual claims are accurate to avoid regulatory non-compliance.

Exact policy details

Responsibility for Accuracy

Organizations must ensure the accuracy of their submissions to the MHRA.

Applies to: All organizations submitting responses to MHRA.

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. · p. 1 · High · Source

Expectations for Submission Accuracy

All submissions/responses must be factually accurate and verifiable.

Applies to: Applicable to all submissions to MHRA Compliance Teams.

All submissions/responses must be: factually accurate and verifiable. · p. 1 · High · Source

Impact of Inaccurate Information

Inaccurate submissions may lead to rejection or request for resubmission.

Applies to: Responses to MHRA inspections.

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. · p. 1 · High · Source

Disclosure of AI Use

Organizations can voluntarily disclose AI use in submissions.

Applies to: Applicable to organizations using AI in responses.

we are offering organisations the option to disclose AI use in responses/submissions to the compliance teams. · p. 1 · High · Source

Workflow rule impacts

Extracted documents

• Use of AI for GXP inspection responses: setting standards without stifling innovation – MHRA Inspectoratehtmlhtml · high · 1 page