Anthropic drops restrictions on AI development in Claude Fable 5
The company is ending a policy that covertly limited Claude Fable 5’s capabilities when users tried to design competing AI systems. Researchers had found that Fable 5 and Mythos 5’s system cards contained a mechanism that intentionally degraded response quality on queries about training pipelines and distributed learning infrastructure.
The decision to lift restrictions came after criticism from the research community, which argued that such measures hindered scientific work and created artificial barriers for independent researchers.
Anthropic has officially confirmed the change in course. The restrictions that were meant to prevent the creation of cutting-edge models with Claude are no longer in effect.
Up next
Anthropic files for IPO
Anthropic confidentially files for IPO. The $1 trillion-valued company eyes a listing as early as fall 2026.
Related stories
- Fable 5 and Mythos 5 models have a hidden mechanism to reduce effectiveness
- Anthropic’s $150M buys Claude time for federal scientists
- Suleyman wants consciousness talk out of Claude’s training
- Anthropic starts counting how much of its R&D AI does
- Anthropic’s alignment lead puts extinction risk above 10%
- Three outside labs ran their own studies on Claude data
Comments
No comments yet. Be the first.
Join the conversation
Sign in to leave a comment. It appears after moderation.