For Lawyerss and Legal Professionals

Protecting Your Judgement: How Lawyersss Can Use AI Without Losing Their Edge

AI legal research tools can hallucinate citations that sound real but do not exist. AI contract templates can embed assumptions you never made. When you outsource the foundational research to AI, you skip the thinking that builds judgement in junior lawyers and protects you from missing risks. The question is not whether to use these tools but how to use them without becoming dependent on their output.

These are suggestions. Your situation will differ. Use what is useful.

Download printable PDF

Verify Every Citation Before You Cite It

Harvey AI, Lexis+ AI, and Westlaw AI produce text that reads like case summaries but can contain invented rulings, wrong court names, or citations to decisions that do not exist. You remain liable for every citation in your memo or brief, regardless of which tool generated it. The tool saves time on the initial search, but you must check each case in the primary source before you rely on it in client advice or court filing. This is not cautious practice. This is mandatory practice.

›Copy the case name and citation into the actual database and read the headnote yourself before using it in any written work
›When a tool summarises a ruling, find the sentence in the judgment itself and verify it matches what the tool said
›Track which citations you had to correct after AI generation so you know which tool has a higher error rate for your practice area

Read the Full Contract Before You Accept the AI Draft

AI contract templates from ChatGPT or specialised tools like Casetext embed legal assumptions based on whoever trained the model. A template for a service agreement might assume unlimited liability caps, short termination periods, or IP ownership rules that do not fit your client's risk tolerance. You cannot just review the sections the AI flagged as important. You need to read every clause and ask why it is there before you send it to your client. The template is a starting point, not a finished product.

›Before you use an AI template, write down the three biggest commercial risks in this deal that the contract must protect against
›Compare the AI template to the last three deals you closed in this category so you can spot what the template assumes differently
›Delete any clause you cannot explain to your client in plain language, even if the template included it

Make Junior Lawyerss Do the Research Before They Use AI Tools

If a junior lawyer starts with ChatGPT or Westlaw AI and skips the basic case reading, they never build the instinct for spotting when a legal argument is weak or risky. The foundational work of reading 10 bad cases before you find the good one teaches you which arguments work and why. If you let AI do that filter, your junior lawyers will not develop the judgement to know when AI has steered them wrong. Require them to spend the first month on a research project reading cases manually before they can speed up their work with AI.

›Create a research checklist that requires junior lawyers to list the weakest case supporting their position before they can use AI to flesh out the memo
›Have them present the case law they found manually before they run it through AI summaries so they build pattern recognition on their own
›Review their unassisted research once a month so they stay sharp even after they start using AI tools for routine work

Do Not Let AI Summarise Risk Analysis

When you ask Lexis+ AI or Harvey AI to summarise the risks in a contract or court ruling, the tool gives you a clean list of top risks. This feels useful. It is not the same as your own analysis. The tool will miss obscure risks that only matter in your specific client context. It will downrank risks that seem small in general but create liability for your client in particular. You need to read the full judgment or contract yourself and come to your own conclusion about what matters. Then you can use the AI summary to make sure you did not miss anything obvious.

›Before you read the AI summary, write your own list of the three to five biggest risks so you are not primed by the tool's ranking
›If the AI summary disagrees with your list, stop and read the source material again to understand why
›In client emails, present risks in your own words drawn from your own reading, not lifted from the AI summary

Keep a Record of Where AI Influenced Your Judgement

If you give advice that turns out wrong, opposing counsel or a regulator may ask what you relied on. If you relied on a Westlaw AI summary without checking the underlying case, you cannot say you conducted proper research. If you used a contract template from ChatGPT without changing it, you cannot say you exercised independent judgement. A simple document trail showing that you reviewed the primary source yourself protects you in a malpractice claim. Write down which AI tool you used for which task and what you verified yourself.

›Add a line to your file notes when you used an AI tool and what you did to verify the output before you relied on it
›Keep screenshots of the AI output you received if a case relies on a particular interpretation the tool provided
›If a junior lawyer used AI research on a matter, have them note in the memo which cases they verified personally

Key principles

1.You are liable for every citation and legal conclusion in your work, whether you generated it yourself or an AI tool did.
2.AI contract templates and summaries encode assumptions that may not match your client's actual risk tolerance or business situation.
3.Foundational legal research done manually is how lawyers develop the judgement to spot when AI has made an error.
4.Risk analysis requires your own reading and reasoning because AI cannot know which risks matter most in your specific client's context.
5.A written record that you verified AI output before relying on it is your best defence if a matter goes wrong.

Key reminders

When Harvey AI or Westlaw AI produces a case citation, open the full judgment yourself before you write anything that cites it
Have junior lawyers present their manual research findings before they use AI to expand or summarise that work
Read contracts and court rulings in full yourself even when AI tools offer to save you time with summaries
Create a simple file note every time you use an AI tool that documents what you did to verify its output
Compare AI summaries of risk to your own analysis in private notes before you present risks to the client