Anthropic officially launched its most advanced ‘Mythos Class’ model, Claude Fable 5, on Tuesday, June 09, 2026. It is making headlines for its invisible safety guardrails, which weren’t declining queries but instead silently degrading model performance by producing lower-quality outputs or incorrect answers. And it was done to queries related to AI research and development.
This led to severe backlash from AI researchers, developers, and scientists, who argued that it creates anti-competitive behavior and harms open research. Majorly in the fields of cybersecurity, biology, chemistry, and AI distillation topics.
Anthropic said that they implemented the restrictions to prevent users from misusing the model
“We wanted to be able to provide this level of intelligence for general users in a safe manner.” Dianne Penn, Anthropic’s head of product management, research, and labs.”
Some key takeaways about Anthropic’s latest Model
- The safety classifiers that decline certain requests in Claude Fable 5 are not applicable in Claude Mythos 5.
- Both Claud Fable 5 and Claude Mythos 5 have the same context window and output, which is a 1M-token context window and up to 128k tokens per request.
- Users can access Claude Fable 5 through the Claude API, Claude Platform on AWS, Amazon Bedrock, Vertex AI, and Microsoft Foundry.
- Claude Mythos 5 is available to a limited number of approved customers in Project Glasswing.
- Both models contain 30-day data retention and are not available under zero data retention.
Why is Microsoft stopping employees from using Claude Fable?
The employees can access every Claude model except for Fable 5, because it comes with data retention, and Microsoft’s legal team wishes to hold the line there. As previous Claude models supported the Zero Data Retention (ZDR), the latest one doesn’t; the legal team of Microsoft is reviewing the retention policies and confirming whether the policies are compliant with safety standards.
After a lot of backlash from individuals and organizations, Anthropic made a statement,
“We’re changing Fable 5’s safeguards for frontier LLM development to make them visible,” Anthropic said in a statement to WIRED. “We made the wrong trade-off, and we apologize for not getting the balance right.”
Also Read: Grok AI vs ChatGPT: Which AI Tool Is Better in 2026?
