Major AI models systematically refuse to criticize restrictive leaders or governments (new study on built-in censorship/bias).

7.0Weirdness

Why It Matters

Needs editorial pass.

Evidence

AP News coverage of the study; HN discussion under ChatGPT/LLM tags.

Signal Read

Novelty: 7Receipts: 7Story voltage: 7Heat: 7

Source Trail

Daily scan: 2026-07-20