The flow moves from source to reviewable work.
Good AI PracticeRecommended operating sequence.
Frame AI use case · Map decision impact · Assess data suitability · Validate model performance · Set human review · Monitor lifecycle · Prepare regulator-facing explanation
Where teams usually lose traceability.
audit watchSource skipped
Using a blog summary instead of the official source creates weak claims.
Role unclear
Provider, deployer, sponsor, manufacturer, vendor, or owner responsibilities blur.
Evidence not owned
A document exists but nobody owns review, update, or closure.
Change not logged
Model, system, SOP, source, or guidance changes do not reach the evidence file.
Current public sources for Good AI Practice.
official firstThese links are the public source anchors for this workspace. Interpretation, checklists, and future assets should point back here before being reused outside iFeed.
FDA guiding principles for Good AI Practice in drug development
2026-01-14 · FDA page for common Good AI Practice principles in drug development.
EMA and FDA common AI principles news
2026-01-14 · EU-facing publication context for the common principles.
EMA/HMA AI in medicinal product lifecycle
2024-09-30 · Broader medicinal-product lifecycle reflection layer for AI use and governance.
FDA/Health Canada/MHRA GMLP principles
2021-10-27 · Ten guiding principles for Good Machine Learning Practice for medical device development.