No Surprises

The first rule I learned in federal grants work: no surprises. If something went wrong — or might go wrong — you told the funder before they found out on their own. Bad news delivered early was a problem to solve together. Bad news discovered late was a breach of trust.

AI needs a no-surprises rule, pointed in every direction.

Tell your customers when AI is involved in decisions about them — before they discover it. Tell your team what the system can and can't do — before they over-trust it. Tell leadership when something goes wrong — before it becomes the kind of story that starts with "we didn't know."

This isn't about confessing every detail. It's about the order of discovery. People forgive problems they hear about from you. They don't forgive problems they uncover themselves.

Grants officers could forgive a struggling program that communicated. They could not forgive being blindsided. Your customers, your staff, and your regulators work exactly the same way.

No surprises. Say it early, say it plainly.

I explain frameworks; I don't give legal counsel.