Communication Templates (AI Safety Pack Rollout)
Version: v1.0 Ready‑to‑adapt announcements for different audiences.
Template 1: Executive Announcement (Leadership/All‑hands)
Subject: New AI Use Policy — Using AI Responsibly at [Org]
To: All staff From: [CEO/Leadership] Date: [Date]
Team,
AI tools are transforming how we work. Used well, they can boost productivity. Used carelessly, they can put people at risk and expose sensitive data.
Today we're publishing our AI Use Policy. It strikes a balance: enabling productive use while protecting against people harm and data leakage.
What you need to know:
✓ Approved uses: Drafting internal documents, brainstorming, code assistance, translating non‑sensitive content
⚠️ Conditional uses (with controls): Customer support drafts, data analysis on confidential info, code copilot — these require human review and logging
✗ Prohibited uses: Making hiring decisions, ranking candidates, pasting customer personal data into unapproved tools, creating deceptive content
Your actions:
- Read the policy (attached / [link])
- Review the Approved/Prohibited 1‑pager
- Complete mandatory training by [date]
- Report any incidents or near‑misses to [channel/email]
Questions? Contact [Risk/Security contact] or your manager.
We believe in using AI responsibly. This policy helps us do that.
[Name] [Title]
Template 2: Manager Briefing (for team leads)
Subject: AI Policy Rollout — Manager Talking Points
To: All managers From: [Risk/HR] Date: [Date]
Managers,
Your teams will receive the AI Use Policy this week. Here's what you need to know to answer questions.
Key messages:
- This isn't about banning AI — it's about using it safely
- The default is "start with the matrix" — not "ask permission for everything"
- Incidents should be reported, not hidden (we're building a learning culture)
Common questions and answers:
Q: "Can I use ChatGPT for work?" A: Only if it's on the approved tools list. Check [link] or ask IT.
Q: "What if I already did something that might be against the policy?" A: Report it as a near‑miss. We need to assess risk, not assign blame.
Q: "The policy says I need human review for customer drafts. What does that mean?" A: Before sending an AI‑drafted message to a customer, a qualified person must check it for accuracy, tone, and data leakage.
Q: "Other companies automate this. Why can't we?" A: Our risk tolerance is documented. If you believe an exception is warranted, we have a formal exception process.
Escalation path:
- Technical questions → IT/Security
- Policy interpretation → Risk/Legal
- Team concerns → HR
Required actions:
- Attend manager briefing session ([date/time])
- Discuss with your team within 48 hours of policy announcement
- Ensure 100% training completion by [date]
Thanks for your leadership on this.
[Name] [Title]
Template 3: Training Invitation
Subject: Mandatory Training: AI Safety at [Org]
To: [All staff / Specific role groups] From: HR / Risk Date: [Date]
You're invited to AI Safety Training
AI tools can boost productivity — but only when used responsibly. This session covers our policy, practical guidance, and how to avoid common risks.
Date: [Date] Time: [Time] [Timezone] Duration: 75 minutes + Q&A Location: [Zoom link / Conference room] Mandatory: Yes (attendance tracked)
What you'll learn:
- What AI is good at (and what it's not)
- Our approved vs prohibited use‑cases
- How to classify a new use‑case using the matrix
- Data handling rules (what you can/can't paste)
- Human review requirements
- How to report incidents and near‑misses
Pre‑work:
- Review the AI Use Policy (attached)
- Review the Approved/Prohibited 1‑pager
Quiz: Short quiz at the end (must pass with 8/10)
Questions? Contact [contact]
[Add to calendar]
Template 4: Incident Response Communication (Internal)
Subject: AI Incident Report — [Brief Description]
To: Risk Committee, Security, [Business Owner] From: [Reporter / Security] Date: [Date] Incident ID: AI‑[YYYY]‑[###]
Summary: [One‑sentence description]
What happened: [Timeline of events]
Impact:
- Data involved: [classes]
- People affected: [number/description]
- Systems affected: [if any]
Immediate actions taken:
- [Containment steps]
- [Notifications made]
Root cause (preliminary): [Initial assessment]
Next steps:
- [Investigation actions]
- [Remediation actions]
- [Communication plan]
Owner: [Name] Next update: [Date/time]
Template 5: Near‑Miss Kudos (Culture Building)
Subject: Near‑Miss Reported — Thank You
To: All staff (or relevant team) From: [Risk/Security] Date: [Date]
Team,
A team member reported a near‑miss this week: [brief, anonymized description, e.g., "almost pasted customer data into an unapproved tool but caught themselves and reported it"].
Why this matters: Near‑miss reporting helps us learn before something serious happens. It takes courage to report, and we appreciate it.
What we learned: [One sentence about the lesson]
What changed: [One sentence about the fix, if any]
Keep reporting: If you see something, say something. Report to [channel/email].
Thank you for keeping us safe.
[Name] [Title]
Template 6: Exception Request Announcement (if approved)
Subject: Approved Exception — [Use‑Case Name]
To: Affected team, Risk Committee From: [Risk Committee Chair] Date: [Date] EDR ID: EDR‑[YYYY]‑[###]
Exception approved: [Use‑case name] has been approved as a time‑boxed exception with compensating controls.
What was requested: [Brief description of what the team wants to do]
Why it was approved: [Business rationale + risk assessment summary]
Conditions:
- Effective: [date]
- Expires: [date] (max 6 months)
- Required controls: [list key controls]
- Monitoring: [metrics and thresholds]
- Kill switch owner: [name]
Your responsibilities: If you're part of this use‑case:
- Complete additional training by [date]
- Follow the workflow exactly as documented
- Report any issues immediately to [contact]
Questions? Contact [Risk contact]
[Name] [Title] Risk Committee Chair
Template 7: Quarterly Update (Ongoing Governance)
Subject: Quarterly AI Safety Update — [Q#] [Year]
To: Leadership, Risk Committee, All Managers From: [Risk/Security] Date: [Date]
This quarter at a glance:
| Metric | Value | Trend | |---|---|---| | Active use‑cases | [#] | ↗ | | Training completion | [%] | → | | Incidents reported | [#] | ↘ | | Near‑misses reported | [#] | ↗ (good!) | | Exception reviews | [#] | → |
Key highlights:
- [New use‑case approved]
- [Control improvement made]
- [Lesson learned from incident/near‑miss]
Focus for next quarter:
- [Planned improvement]
- [Upcoming review]
Open items:
- [Exception expiring — renewal decision needed]
- [New tool evaluation]
Questions? Contact [contact]
[Name] [Title]