AI is changing how in-house legal teams operate. Used well, it can save hours, reduce errors, and free lawyers up to focus on strategic work. But used badly, it can waste time, create risk, and damage your credibility.

Here are seven mistakes I see in-house lawyers make with AI and how to avoid them.

1. Treating It Like Google

AI doesn’t just find information. It generates new material based on your instructions. Treat it like a junior lawyer - you need to give context, detail, and direction.

2. Relying Only on Generic Tools

Practising with ChatGPT is fine, but it’s not built for in-house legal workflows. Tools designed for lawyers are more reliable, secure, and better integrated with the places you already work (like Slack and email).

3. Not Being Specific

If you ask for “key risks,” you might get three or thirty. Be precise: “List five potential risks in this contract.” AI performs best when you set clear boundaries.

4. Withholding Documents

AI works better when it has the right inputs. Upload the contract, policy, or case law you’re working on instead of summarising it yourself. This reduces errors and makes outputs more useful.

5. Asking for Answers Instead of Explanations

Jumping straight to “what’s the answer?” can produce oversimplified or misleading results. Ask AI to explain its reasoning first, then refine toward an answer.

6. Skipping the Persona

AI is a copycat. If you don’t give it a role, you’ll get generic output. Define the persona you want — for example: “You are an in-house counsel at a scaling SaaS company. Analyse this clause for compliance risks and give me five talking points for my leadership team.”

7. Stopping After the First Draft

AI improves with iteration. Don’t settle for the first answer. Keep adjusting your prompts and treating it like a conversation.

Bottom line: AI can be a powerful co-pilot for in-house lawyers, but only if you use it with intention. Treat it like a colleague who needs clear instructions, the right context, and a bit of back-and-forth.