Section 059 · Chapter 8, Operating AI: Observability, Relevance, and Economics

Data Contracts for AI Systems

AI systems need explicit contracts for what they receive, produce, cite, log, refuse, and do.

data contractrefusaldata contracts

What to do

  1. Start with input contracts.
  2. Define required fields, allowed formats, maximum sizes, language assumptions, privacy classifications, and what happens when data is missing or malformed.
  3. Define prompt contracts.
  4. Define tool-call contracts.
  5. Define output contracts.

Evidence to preserve

  • Preserve the inputs, versions, configurations, raw outcomes, and results for data contract, refusal, data contracts needed to reproduce work on Data Contracts for AI Systems.
  • Report results for data contract, refusal, data contracts by relevant slice, separate blocker failures from averages, state uncertainty and blind spots, and connect the result to a release decision.

Expert note

In production work, AI data contracts should be machine-validated, versioned, attached to traces, enforced at runtime, and tested with malformed inputs, adversarial prompts, missing fields, tool errors, and policy changes.

Continue the conversation

Apply this to your context.

Save your product context once, then open a focused conversation that combines it with this concept.

Cite this page

Jason Arbon. "Data Contracts for AI Systems." Testing AI Knowledge Edition, section 59.

https://jarbon.ai/testing-ai/knowledge/ch059-data-contracts.html

Shared across the Knowledge Edition

Adapt every concept to your world.

Describe your product, role, users, risks, constraints, or current quality problem. This stays in this browser until you choose to send it to ChatGPT.

Saved only in this browser.0 / 2400