Checklist: AI Legal Research

Know Your Tool's Limitations

• Confirm your platform's training cutoff date and treat anything that postdates it as potentially missing from your results.

• Find out whether your platform is connected to a curated legal research database or drawing from publicly available web content — the difference determines how much you can trust the coverage.
• If your platform has web browsing enabled, understand that web-retrieved results aren't vetted the same way as content from a legal database.
• Remember that AI retrieval is prompt-dependent — it returns results responsive to what you asked, not everything relevant to your argument; a product liability case with a directly applicable negligence analysis won't surface if you searched for slip-and-fall cases.
• Be especially cautious in areas where AI coverage is thinnest: recent decisions, state and local case law, administrative agency guidance, regulatory decisions, and local rules and trial court orders.

Protect Your Research

• Treat every AI research output as a starting point, not a finished product. Use it to get oriented, then follow up with manual research and verification.

• Run a final check through your research tool before you're done, particularly on fast-moving areas of law where recent developments are most likely to matter.
• Verify that cases your AI surfaced have not been appealed, distinguished, or overturned since the platform's training cutoff.
• Remember that incomplete research doesn't raise an AI ethics question. It raises a competence one; under the ABA Model Rules, competence includes understanding the limitations of the technology you use.