Anthropic · Claude · Mythos · Apple · Data Center · Claude Code · The Register
It blocked us at 'hello!' Anthropic Fable 5 refusing innocuous prompts
Compiled by KHAO Editorial — aggregated from 1 source. See llms.txt for citation guidance.
◌ Single Source
Hyper-vigilant safety classifiers turn Fable into cautionary tale.
Key facts
- These include: [Bug] Fable 5 model safety filters causing false positives on benign messages #66587; Fable 5 refuses to assist with 'Application Security Architect resume' editing #66655
- Mike Famulare, principal research scientist at the Institute for Disease Modeling, part of the Global Health Division of the Gates Foundation, reports ( #66657 ) that Claude Fable 5 balks at inputs
- As developer Clay Merritt fumes, "Anthropic’s Fable 5 silently sabotages its answers when it detects AI/ML work
- But with an estimated 18 to 30 million users worldwide, even a small percentage of thwarted users makes a racket
Summary
UPDATED Anthropic's newly released Claude Fable 5 generative AI model is trying so hard to be safe that it's hurting its own userbase. Anthropic warned that it had tuned Fable 5's guardrails conservatively: "They’ll sometimes catch harmless requests, though they trigger, on average, in less than five percent of sessions," the company said, promising to "reduce false positives as quickly as they can. Anthropic spins a Fable of a tamer, safer Mythos. GM gets datacenter fever, decides to build grid-scale sodium-ion batteries. The company did not immediately respond to a request to quantify model refusals.