Course Content
Prompt Engineering for Legal Applications

Legal Use Cases & Risk Tiers

Not every use case has the same risk. The key is to match the tool to the task and add controls where the stakes are high.

{{UPLOAD_ASSET:llm_usecase_risk_matrix.png}}

Matrix of legal AI use cases by risk and value
Use-case risk matrix: low-risk drafting vs high-risk legal conclusions (conceptual).

Key takeaways

  • Match the tool to the task: higher stakes require more controls.
  • Treat citations as untrusted until verified.
  • Avoid putting sensitive client data into public AI tools.

Low-risk, high-value tasks

  • Drafting internal outlines
  • Summarizing a provided document (you supply the text)
  • Creating checklists and workflows

Medium-risk tasks (require verification)

  • Drafting a client email (must review facts and tone)
  • Generating a first-pass issue list for a motion
  • Spotting missing clauses in a contract (still verify)

High-risk tasks (do not delegate judgment)

  • Final legal conclusions
  • Novel or unsettled legal questions
  • Anything involving confidential client data in an unvetted system

Your risk controls toolkit

In later modules, you’ll learn controls such as: jurisdiction constraints, required citations, explicit uncertainty, step-by-step checklists, and external verification.