AI Governance

A policy won’t change AI behaviour. Confidence will

AI policies alone don’t change behaviour, confidence does. In this article, Martin Adams explores why most AI policies fail in practice, the fears and uncertainty many employees have about AI, and the practical steps leaders can take to build confidence so safe AI use becomes part of everyday work.

Published on LinkedIn • 5 Jul 2026

Read on LinkedIn

Most AI policies fail for one simple reason: people don’t remember them when they’re under pressure.

When it comes to AI, plenty of people are nervous. Some are excited. Some are quietly worried about what AI means for their job. If you don’t talk about that last part, you’ll get one of two outcomes:

people avoid AI completely (and they and your organisation fall behind), or
people use it quietly (and you lose visibility and control)

What works best is regular, short, practical guidance. Not once a year. Not a PDF nobody reads.

Do this instead:

Share a quick but relevant example every few weeks: “Is this safe use of AI? If not, what’s the safe alternative?”
Repeat the one rule that matters most (as an example, yours might be “no sensitive data into unapproved tools”)
Make it safe to ask. If someone checks before they paste something in, that’s exactly the behaviour you want.
Invite questions and feedback – you can’t address fears you haven’t heard

And say it plainly: AI is here to help people do better work and enable them to focus on the value they can add, not to catch them out.

Key question: what’s the bigger blocker in your organisation – people don’t have AI skills, or people worried about what AI means for them? Once you are honest about the answer to this question, you can start to address the issue.

Tagged AI, Governance

A policy won’t change AI behaviour. Confidence will

Most AI policies fail for one simple reason: people don’t remember them when they’re under pressure.

When it comes to AI, plenty of people are nervous. Some are excited. Some are quietly worried about what AI means for their job. If you don’t talk about that last part, you’ll get one of two outcomes:

people avoid AI completely (and they and your organisation fall behind), or

people use it quietly (and you lose visibility and control)

What works best is regular, short, practical guidance. Not once a year. Not a PDF nobody reads.

Do this instead:

Share a quick but relevant example every few weeks: “Is this safe use of AI? If not, what’s the safe alternative?”

Repeat the one rule that matters most (as an example, yours might be “no sensitive data into unapproved tools”)

Make it safe to ask. If someone checks before they paste something in, that’s exactly the behaviour you want.

Invite questions and feedback – you can’t address fears you haven’t heard

And say it plainly: AI is here to help people do better work and enable them to focus on the value they can add, not to catch them out.

Key question: what’s the bigger blocker in your organisation – people don’t have AI skills, or people worried about what AI means for them? Once you are honest about the answer to this question, you can start to address the issue.