AWS AI Service Cards
AWS's transparency docs for responsible AI use.
❓ What is it?
Published documents in which AWS explains, per AI service/model, the intended use cases, limitations, responsible-AI design choices, and deployment best practices — e.g. cards for Amazon Titan, Rekognition face matching, and Textract.
💡 Why does it exist?
Responsible AI needs transparency from the provider, not just the builder. Service Cards let you assess whether a service's known limitations are acceptable for YOUR use case before you ship it.
⏱️ When should you use it?
Read the relevant card during service selection and risk assessment — especially for high-stakes uses (biometrics, lending, hiring) where documented limitations drive design safeguards.
🗺️ Where does it fit?
In your governance process rather than your architecture: they inform model risk assessments, DPIAs, and the human-oversight controls you attach around a service.
🔌 How do you integrate it?
Find the card on the AWS Responsible AI site, map its stated limitations to your use case, and document mitigations (thresholds, human review, monitoring) in your model governance records.
🧩 Commonly integrated with
🎯 Exam angle (AIF-C01)
- "Where does AWS document intended use and limitations of its AI services?" → AI Service Cards — a favourite responsible-AI question.
- Your equivalent for models you build = SageMaker Model Cards; AWS's equivalent for their services = AI Service Cards.