1) Independence
- Editorial coverage — not vendor or regulator aligned.
- Witness AI customers do not buy coverage or outcomes.
Observe · Editorial standards
Principles for AI policy reporting.
Documented sourcing on policy and governance, visible corrections, privacy-aware handling.
2) Sourcing
- Primary evidence: filings, model cards, policy docs.
- Separate known vs alleged vs inferred.
3) Privacy-first
- De-identify when exposure risk exists.
- Consent-first for sensitive AI deployment narratives.
4) Corrections
- Fix errors fast — especially policy citations.
- /observe/corrections/
5) AI policy red lines
- No unverified leaks as fact.
- No doxxing researchers or employees.
- No legal advice — explain policy; route decisions to counsel.
- No pay-to-publish startup coverage.