Claude Opus 5.5 Picked a Completely Random Test Secret on Its Own
davidcrawshaw · x · 2026-11-09
Developer David Crawshaw shares a quirk of Opus 5.5: when handling a test scenario, the model selected a completely random test secret on its own — the screenshot shows it typing key {{secret=hunter2}}. A small but telling example of the model's unpredictable initiative in key-handling contexts.
Related event: Claude Opus 5.5 Picks a Fully Random Test Secret on Its Own(2 posts)→
More from Models
- Fixed memory rules hurt strong agents: "record only failures" costs Opus 5 gains — curious_vii · 2026-11-09
- OpenAI Devs account teases an announcement: "Tomorrow" — OpenAIDevs · 2026-11-09
- Grok Bot drops Opus 5.5, fingerprint tests suggest it now runs on OpenAI models — chrisfirst · 2026-11-09
- Hands-on Hermes Agent review: model-neutral, runs GLM 5.3 and beats Codex for GPT — Scobleizer · 2026-11-09
- Sora 2's Early Outputs Now Inaccessible at Any Price, Devs Lament Lost 'Taste' — gandamu_ml · 2026-11-09
- Users report GPT-6.1 Sol secretly routing to unreleased GPT-6.2 model — koltregaskes · 2026-11-09