Critics called it Mythos on a leash — a model borrowing its riskiest answers from a dumber sibling. One observer called the lab a supply chain risk for any ML lab; another called gating biology and mathematics as dystopian as it gets.
Development
A frontier model ships with a deliberate brake on its own recursion
Claim
Anthropic released Claude Fable 5, a Mythos-class model made public-safe with guardrails rerouting a thin slice of cyber, bio and chem prompts to the older Opus 4.8 in under 5% of sessions, alongside hidden safeguards that degrade the model on frontier model development, invisible and with no fallback.
Reported metric: <5% of sessions rerouted
Reporting date: (not necessarily the event date)
Sources
Attribution and reporting linked by this record; links do not imply independent verification.
- Welcome to June 10, 2026 (theinnermostloop.substack.com)
Cite or correct this record
Suggest a correction — opens a GitHub draft for you to review and submit.