AI for Legal and Compliance Teams
In-house counsel, compliance officers, contract managers and paralegals who review, negotiate and summarise documents for a living
Prerequisites: Module 101 (AI at Work 101) or your platform's 101 module (01 Claude 101, 201 Gemini 101, 301 ChatGPT 101, 401 Copilot 101), and Module 02 (How to Prompt)
Last verified 2026-09-17
Learning objectives
- Extract clauses from a contract with verbatim quotes and locations
- Compare a contract to a playbook and get findings you can check in the document
- Write plain-language summaries that preserve obligations
- Maintain a clause library the assistant drafts from
- Apply the four guardrails, including the citation rule: no authority is real until you have opened it
- Decide, in advance, what never goes into an assistant on any plan
What's in this module
6 lessons · 5 exercises · ~8 min reading
Starts withVerbatim or nothing~1 minExecutive summary
You will learn three contract-review patterns (clause extraction, playbook comparison, plain-language summary), how to build and use a clause library, and the four guardrails that decide whether any of it is safe: confidentiality, privilege, hallucinated citations, and what never leaves the firm. End state is one contract extracted and compared against your playbook with every finding quoted, one plain-language summary a business reader signed off, and a written list of what your team does not put into an assistant. After this module you can do this on Claude, Gemini, ChatGPT or Copilot, and explain to a partner or a regulator where the line is and why.
Sources
- Anthropic, "Reduce hallucinations", quote-first extraction, permission to say "I don't know", and citation verification: https://platform.claude.com/docs/en/test-and-evaluate/strengthen-guardrails/reduce-hallucinations
- Anthropic Privacy Center, "Is my data used for model training?", the commercial-plan exclusion: https://privacy.claude.com/en/articles/7996868-is-my-data-used-for-model-training
- Google Workspace Admin Help, "Generative AI in Google Workspace Privacy Hub", the Workspace commitment on training and human review: https://knowledge.workspace.google.com/admin/gemini/generative-ai-in-google-workspace-privacy-hub
- OpenAI Help Center, "Data Controls FAQ", the personal-plan training toggle and the business-plan controls: https://help.openai.com/en/articles/7730893-data-controls-faq
- Microsoft Learn, "Data, privacy, and security for Microsoft Copilot", no training on tenant data, permission inheritance and sensitivity labels: https://learn.microsoft.com/en-us/copilot/microsoft-365/microsoft-365-copilot-privacy
Prefer the live room?
This module also runs inside our in-person bootcamps and workshops in Copenhagen.
More in Track H
AI for Sales and Revenue Teams
The five sales workflows where an assistant pays for itself: pipeline triage, account and meeting prep, proposal and follow-up drafting from a brief, objection practice, and forecast commentary.
AI for HR and People
How to turn a role brief into a job ad and an interview kit, draft and plain-language a policy, build onboarding material, and run a bias check on all of it, without ever pasting a candidate or employee record into the wrong place.
AI for Finance
A verification discipline for finance work with an assistant: assumptions before formulas, formulas before numbers, a reconciliation check before anything leaves your desk, and commentary that points at the cells it came from.