Oliver is a junior solicitor at a busy law firm. Sharp, ambitious and quietly using AI to draft client documents, summarise case notes and speed up his workload. Nobody had set any rules.
Last week a senior partner reviewed one of his briefs.
The AI had confidently included case references that don’t exist. Cited legislation that had been repealed. And summarised a client’s position in a way that was plausible, fluent and completely wrong.
The brief had already gone to the client.
Now the firm has a potential negligence issue and a client who is asking questions.
And a junior solicitor who genuinely didn’t know he was doing anything wrong.
Because nobody told him what AI could and couldn’t be used for. There was no policy. No approval process. No training. Just a bright young lawyer and a tool that sounds authoritative even when it’s wrong.
✅ 3 THINGS EVERY FIRM NEEDS BEFORE AI TOUCHES CLIENT WORK
1. An AI acceptable use policy — which tools are approved, what they can be used for and who signs off the output.
2. Mandatory human review of any AI-generated content before it leaves the building — especially anything client-facing.
3. Training on AI limitations — confident output is not the same as accurate output, and every member of staff needs to understand the difference
At People Vision HR we help businesses build clear, practical AI use policies and deliver training that means the next Oliver knows exactly where the line is.
Before a client brief becomes a negligence claim.