Anthropic has redeployed its Fable 5 model after a temporary suspension due to a discovered bypass vulnerability. While Mythos 5 remains restricted, Fable 5 is now globally available, with initial usage offered at a discount. The vulnerability, which allowed bypassing safeguards, was found to affect other leading models as well, including Claude Opus 4.8 and GPT-5.5. Anthropic has implemented a new classifier to block such bypasses in over 99% of cases, though this may lead to an increase in false positives. AI
IMPACT Highlights the fragility of deploying frontier models and the impact of policy on product timelines, encouraging consideration of self-hosted alternatives.
RANK_REASON Frontier-lab model release with system card [lever_c_demoted from frontier_release: ic=1 ai=1.0]
Read on dev.to — Anthropic tag →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
