Ethical Guidelines and Human Oversight › Communicating Governance to Users
Michael BrenndoerferAugust 16, 20252 min read
AI Agent HandbookAll chapters Part 13 of 15Safety and Governance - Keeping the Agent Aligned
Ethical Guidelines and Human Oversight › Communicating Governance to UsersOne section · Sections in this chapter9 / 12 ↓
Sections in this chapter12
If your agent serves multiple users or is deployed publicly, you should communicate your governance approach. This builds trust and sets expectations.
Here's what to communicate:
What ethical principles guide the agent: Users should know what values the agent upholds.
What oversight is in place: Users should understand when humans review decisions.
How to raise concerns: Users should know how to report problems or ethical issues.
How governance evolves: Users should know that you're actively maintaining and improving the agent's ethical behavior.
For our personal assistant, this might be a simple document:
## Personal Assistant Governance
## Our Ethical Principles
This assistant is designed to:
- **Respect your autonomy**: It asks before making important decisions
- **Protect your privacy**: It never shares your information without permission
- **Treat everyone fairly**: It doesn't discriminate or make biased assumptions
- **Be transparent**: It explains its actions and reasoning
- **Consider impact**: It thinks about how actions affect others
## Human Oversight
- **High-risk actions** (external emails, canceling meetings): Require your approval
- **Medium-risk actions** (internal emails, scheduling): Monitored, you can intervene
- **Low-risk actions** (reminders, queries): Logged for review
## Raising Concerns
If the assistant does something inappropriate:
1. Review the audit log to see what happened
2. Adjust the oversight settings if needed
3. Update the ethical guidelines
4. Report serious issues to [contact]
## Continuous Improvement
We review the assistant's behavior weekly and update its guidelines as needed.
Your feedback helps us improve.