The AI Trust & Safety Check helps teams discuss the human conditions that determine whether responsible AI use is possible. If people hide usage, quality concerns, or mistakes, leaders lose the signals needed to improve the system.
Create honest team agreements
Ask where people feel safe sharing AI use, where trust breaks down, which quality or privacy risks appear, and what agreements would make responsible work easier. Anonymous input can surface gaps that a policy announcement will miss.
End with a short team agreement: what must be disclosed, which data must not enter a tool, when human review is mandatory, how incidents are escalated, and how the team learns without blame.
How to facilitate
- 1Make disclosure safe. Do not ask for honesty and then punish the first person who names a mistake.
- 2Discuss concrete scenarios. Use recent tasks, not abstract ethics language.
- 3Separate learning from misconduct. Define clear red lines while preserving room to report ordinary failures.
- 4Write four usable rules. A short team agreement beats a policy nobody can apply.
When to use this template
Use this check after introducing an AI policy, when usage is hidden, after a quality or privacy incident, or when managers and employees have different expectations about acceptable AI use.
