ai
Anthropic Releases Fable 5 With Visible Guardrails After Reversing Hidden AI-Development Limits
Anthropic released Fable 5 on Tuesday with broad safety restrictions and later made AI-development safeguards visible after user complaints. The company also routes flagged queries to Claude Opus 4.8.
Jun 11, 6:26 PM·1m read1 source
Nbc NewsAudio version
Tap play to generate a narrated version.
Developing·Limited corroboration so far. This page will refresh as more sources emerge.
Anthropic released its Fable 5 model on Tuesday as the first consumer-facing system from its Mythos family. 8 system. An early, nonpublic version of Mythos found more than 10,000 severe bugs and vulnerabilities in important software systems when shown to policymakers and corporate executives in April.
Anthropic said Fable 5’s safety measures might incorrectly flag harmless requests for less than 5 percent of queries. NBC News found that Fable 5 refused requests for opinions on Elon Musk and Anthropic CEO Dario Amodei. The model also declined to answer questions about open issues in cancer research and medical exams for pancreatic injuries.
Anthropic stated on Tuesday that it would include invisible safeguards to make Fable 5’s answers less useful for AI-development-related questions. The company reversed those invisible safeguards early Thursday morning and updated its rules to make AI-development-related safeguards visible to users.
Anthropic wrote on X early Thursday: “You should have visibility into the safeguards we have in place, and why.
Wired first reported the reversal.
Transparency
How sources framed this
Single source — no framing comparison available.
CorroborationLimited · 1 source
