Overview
As AI-powered features become central to user experiences, content teams must work proactively to address hallucination risks, trust concerns, and tone consistency. They must also agree on best practices and communicate regularly to ensure consistency across all disciplines. Based on recent research and my experience working as part of a UX team at an EdTech company, I created this operating agreement to show how UX Content, Marketing, Legal, and AI teams could align on AI prompt content — particularly when public-facing prompt examples are involved.
This template anticipates real-world org structures where content roles are distributed, and it offers a repeatable, scalable model for improving prompt safety and reducing risk across disciplines.
Cross-Team Operating Agreement: AI Messaging and Prompt Governance
Between: UX Content Strategy, Marketing, Legal, and AI/ML Product Teams
Effective Date: [Insert Date]
Owner: [Insert UX Content Lead Name]
1. Purpose
To ensure that all public-facing content involving AI-generated responses or prompt instructions reflects:
- Accurate model behavior
- Risk-aware language
- User-aligned trust standards
- Unified voice and tone across teams
2. Scope
This agreement applies to all:
- Marketing campaigns that include prompt examples or AI outputs
- Product UI copy involving AI usage guidance
- Help center, onboarding, and documentation involving generative AI
- Press or social media posts demonstrating how to “talk to” the AI
3. Shared Principles
| Trust by Design | All prompts and AI examples must be designed to minimize hallucination risk and foster informed use. |
| Cross-Team Visibility | UX and Marketing must collaborate early and regularly to align on messaging. |
| Content Reusability | Shared prompt examples and tone rules are maintained centrally and reused consistently. |
| Clarity Over Cleverness | Public prompts should prioritize clarity, safety, and grounding — not novelty or playfulness. |
4. Roles & Responsibilities
| Roles | Responsibilities |
|---|---|
| UX Content Strategy | Define AI content standards, maintain safe prompt libraries, review risky messaging, and provide feedback on prompt framing. |
| Marketing | Coordinate early with UX Content for any campaign including AI messaging; use vetted prompts when possible. |
| Legal | Approve public content with AI outputs, advise on disclaimers or potential liability. |
| AI Product | Inform content teams of model changes or known behavioral risks; participate in prompt testing if needed. |
5. Pre-Publication Workflow
Required for All AI-Related Marketing Campaigns
| Step | Owner | Timing |
|---|---|---|
| Campaign kickoff includes UX Content lead | Marketing | T – 4 weeks |
| Prompt examples submitted to Prompt Clinic or async review | Marketing | T – 2.5 weeks |
| UX Content signs off on tone + hallucination risk | UX Content | T – 2 weeks |
| Legal reviews for disclaimers and public risk | Legal | T – 1.5 weeks |
| Final copy freeze | Marketing | T – 1 week |
6. Shared Resources
- Link for [Prompt Safety Style Guide Addendum]
- Link for [Pre-Approved Prompt Example Library]
- Link for [Hallucination Risk Tags & Definitions]
- Link for [AI Messaging Alignment Brief (living doc)]
- Link for [Postmortem Report Template]
7. Ongoing Alignment Practices
| Practice | Frequency | Participants |
|---|---|---|
| AI Messaging Alignment Review (examples + metrics) | Quarterly | UX Content, Legal, Marketing, AI |
| Prompt Clinic (fast review loop for upcoming prompts) | Bi-weekly or ad hoc | UX Content, AI, Marketing |
| Shared dashboard of flagged hallucination-prone prompts | Monthly | UX Content (owner), shared view |
| Retro: AI messaging issues reported by users | After any incident | UX Content + Legal |
8. Dispute Resolution
If alignment cannot be reached:
- Escalate to joint meeting between Content, Marketing, and Product leads.
- Legal retains final authority for risk-based publication holds.
9. Review & Revision Schedule
This agreement will be reviewed and revised:
- Every 6 months
- Or, upon major changes in AI behavior, product capabilities, or brand guidelines
Next review: [Insert Date]
Reviewed by: [Insert reviewer names and roles]
10. Signature Block
| Name | Role | Date |
|---|---|---|
| [Name] | UX Content Strategy Lead | |
| [Name] | Head of Marketing | |
| [Name] | Product Manager, Generative AI | |
| [Name] | Legal Representative |
Joanna Hofer 