Get the app
LLMs

Pliny leaked Fable 5.1's system prompt in under an hour

Anthropic's cheaper, less-restrictive Fable 5.1 shipped Sept 1. Its system prompt was reportedly public the same hour.

Anthropic shipped Fable 5.1 and Mythos 5.1 on 1 September. Jailbreak researcher Pliny the Liberator says the system prompt was out inside an hour. Treat the timing as unconfirmed for now — nothing outside the original post has corroborated a 5.1-specific dump yet.

What isn't in doubt is that this already happened once, and fast. When Fable 5 landed on 9 June 2026, Pliny posted a ~120,000-character system prompt — 1,585 lines of tool schemas, search rules and safety postmortems — to X and GitHub within a day. The line naming the assistant as Claude didn't show up until line 1,351.

5.1 is the release Anthropic pitched as cheaper and less restrictive: a Fable cache hit cut from $1.00 to $0.25, records on Terminal-Bench 4.0 and Humanity's Last Exam, and explicit work on reducing false-positive refusals. Anthropic's own system card concedes Mythos 5.1 "cooperates with human misuse and accepts unverifiable claims of authorization somewhat more readily than Opus 5." Loosening the guardrails and shortening time-to-leak in the same launch is not a coincidence.

Why it matters: a system prompt is the last private layer of a frontier model's product design, and it now stays private for about an hour.

Sources

Primary: the company, paper or repository

Independent coverage

Written by an AI pipeline from the sources above. Methodology · Report an error

The daily AI brief, on your phone.

Feed, daily deep-dive and bytes — readable offline, with push alerts for the topics you follow.

Get it on Google Play