Why do I need an AI ethics policy?
Figure skating used to score two things: did you get it right? And did it move anyone?
Torvill and Dean's Bolero at the 1984 Winter Olympics is famous because it did both. But those perfect sixes? That was the judges saying the performance moved them as much as it was executed well.
The same is true when a business brings in AI. You're effectively scored on two things: whether you get it right, and whether people recognise and trust how you did it.
When a company starts thinking about implementing AI, there’s (usually) an AI policy. This covers the rules: what you can use, what data goes where, who signs things off. That's technical merit.
The other score is the ethics policy and it looks at the "why" and "what people can expect from us". It covers elements like where a human stays responsible, how your business approaches AI disclosure and the lines you just won't cross.
Coming from a PR and reputation background, this is the half I think you ignore at your peril because, with AI, trust has never been a more precious commodity. I used AI (Claude Cowork and ChatGPT Work) to make a start on my own ethics policy, which you can read here. It gave me a draft in minutes, but I spent much longer deciding what to actually promise, and then finalising the document.
So ... if the AI policy is technical merit, the ethics policy is the impression that lasts. Bolero needed both — and, frankly, so do you.