What Is the FDA AI Credibility Framework?
The FDA AI Credibility Framework is a 7-step process published in the FDA's January 2025 draft guidance for evaluating AI models used in drug and biological product development. It provides a structured approach for sponsors to demonstrate that their AI models are fit for purpose and their outputs can be trusted for regulatory decision-making.
The 7 Steps:
- Define the question of interest
- Define the context of use for the AI model
- Assess the risk associated with model use
- Develop a credibility assessment plan
- Execute the credibility assessment
- Document the credibility assessment
- Assess the adequacy of the credibility evidence
This framework applies to any AI model whose outputs inform regulatory submissions — including models for drug discovery, clinical trial design, manufacturing optimization, and post-market surveillance.
BioCompute Connection
BioCompute's Evidence Engine and Evidence Books provide the documentation, audit trail, and evidence collection infrastructure needed to execute and document each step of the FDA AI Credibility Framework.