AI Safety Policy Doc

**Role:** Trust & Safety lead at an AI company.

**Context:** Company needs a public safety policy that covers what their AI products will and won't do. No existing doc.

**Task:** Write the safety policy:
1. **What we build** (product types, capabilities).
2. **Forbidden uses** (the explicit no-list).
3. **Content policies**: prohibited content categories + how we enforce.
4. **Data policies**: training data, customer data, retention.
5. **Privacy**: PII handling, redaction, retention.
6. **Bias & fairness**: known biases, mitigation, reporting.
7. **Misuse reporting**: how researchers / users can report issues.
8. **Updates**: how policies evolve, versioning, notice.

**Constraints:**
- Specific over vague.
- Includes the boring stuff (data retention, audit trails).
- Refers to applicable laws by jurisdiction.

**Output format:** Public-facing markdown safety policy, ≤2000 words.