An employee is coming to see you about a restructure, a performance review, a declined leave request.
Before they walk in, they’ve described the whole situation to an AI and asked whether what you’re doing is fair.
That’s not a prediction. New Zealand employment specialists at Peninsula flagged it in September: staff are using AI as an informal employment-law adviser, and arriving at meetings with AI-generated views on their rights.
The wrong reaction
The easy reaction is to treat this as a threat, and I think that reaction is a bit of a tell.
What has actually happened is that an information gap closed. For most of working life the employer had faster access to precedent, policy and legal context than the employee did. Not because employers are sinister — because they had the advisers, the handbook and the history. That asymmetry is gone now, for anything a language model can reason about.
Here is the uncomfortable version, and I’d rather write it plainly than dress it up:
Your management decisions are now being reviewed for consistency, by something with infinite patience and no fear of you.
If your decisions hold up, almost nothing changes, except that people arrive at conversations better informed — which, on balance, is good. If they only held up because nobody ever checked, that’s not an AI problem. That’s a problem AI found.
What I’m not saying
I want to be clear about this, because there’s a version of this article that’s ugly and I’ve seen its cousins.
I’m not saying your staff are building cases against you, and I’d be wary of anyone selling you that anxiety. Most people doing this are doing exactly what people have always done before a hard conversation: trying to understand it. They used to ask a mate, or a family member who once worked in HR, or read something on a union website. The advice is now faster, better sourced and available at eleven at night when they can’t sleep.
The behaviour hasn’t changed. The quality of the second opinion has.
The gap nobody’s covered
Peninsula’s sharpest observation is about policy. A lot of organisations have now written something about employees using AI for work. Fewer have thought about employees using AI about work.
That’s a real gap, and it’s not one you close with a rule — you can’t sensibly tell someone not to think about their own employment with the tools available to them, and you’d look absurd trying.
What you can do is stop being surprised by it.
Three things I’d actually do
Assume it. Walk into every significant conversation assuming the other person has already run it past an AI and formed a view. Practically, this means being able to explain your reasoning, not just announce your decision. “Because that’s the policy” was always a weak answer. It’s now a visibly weak answer.
Treat consistency as an asset. The single thing an AI is best at spotting is a decision that doesn’t match the one you made for someone else six months ago. If you’ve been inconsistent, you’ll find out. Better to audit that yourself first.
Say it out loud. If you’re genuinely comfortable with your people researching workplace matters — and you should be — tell them. The alternative is a team that assumes you’re not, which is one more thing they feel they have to hide from you. Most businesses already have enough of those.
The part that was always true
The conclusion I’d underline cuts against the reflex to buy something.
The organisations that come through this well won’t be the ones that invested most in AI. They’ll be the ones whose management decisions hold up when someone checks them, and whose people believe they can raise something without it becoming a fight.
That was always true. It’s just now being tested, continuously, by a third participant who never gets tired and never takes a hint.