Anthropic reactivates Fable, Mythos after securing government approval

2026-07-01 · Source: Information and Enterprise Technology News | CIO Dive - Www.ciodive.com · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cybersecurity & Data Privacy, Emerging Technologies & Innovation · Depth: Fundamental Awareness, short

Summary

Anthropic reactivated its Fable and Mythos AI models on July 1, 2026, following the Trump administration's lifting of an export-control ban. The ban, initiated by the Commerce Department after Amazon identified a potential circumvention of Fable's safeguards, prompted Anthropic to develop an "improved safety classifier." This new safeguard, tested and approved by NIST's Center for AI Standards and Innovation, aims to block reported bypass behaviors, though it may increase false positives for cybersecurity researchers. Fable 5 is now generally available, while the more powerful Mythos 5 remains limited to trusted partners. The company is also addressing broader industry concerns about the government's vetting process for frontier AI models, committing to provide early access for national security-relevant models and share threat intelligence. Anthropic is collaborating with Amazon, Google, and Microsoft on Project Glasswing to establish a "consensus framework" for classifying AI jailbreak severity, based on four criteria, to standardize model review.

Key takeaway

For AI Security Engineers evaluating frontier models, Anthropic's experience highlights the need for robust, independently verified safeguards. You should prioritize models with transparent safety classifiers, understanding their potential for false positives in defensive tasks. Actively participate in or monitor initiatives like Project Glasswing to adopt standardized jailbreak classification frameworks, which will streamline your internal model vetting and risk assessment processes.

Key insights

AI model deployment requires robust safeguards and standardized vetting processes, especially for frontier capabilities.

Principles

• AI safety classifiers can introduce false positives.
• Standardized jailbreak classification is crucial for model review.
• Government-industry collaboration is key for responsible AI deployment.

Method

Anthropic trained an improved safety classifier to block reported bypass behaviors, tested by NIST. It also works with Project Glasswing partners to develop a four-criteria framework for jailbreak classification.

In practice

• Implement improved safety classifiers for AI models.
• Collaborate on industry-wide jailbreak classification standards.
• Engage government for early access and threat intelligence sharing.

Topics

• AI Export Controls
• Frontier AI Models
• AI Safety Classifiers
• AI Jailbreak Classification
• Government AI Policy
• Project Glasswing

Best for: CTO, VP of Engineering/Data, Investor, Policy Maker, AI Security Engineer, Director of AI/ML

Related on AIssential

• After spooking Trump into safety testing, Anthropic AI models get global release — AI model export controls can be lifted through enhanced government collaboration and robust safety protocols.
• Anthropic's Fable 5 is back worldwide after a two-week government ban over a jailbreak — AI models, even with enhanced guardrails, remain inherently vulnerable to jailbreaks, necessitating continuous security evolution and industry standards.
• "They screwed us": Personality clashes sent Anthropic's models offline — Personality clashes and jailbreak concerns prompted a US government directive suspending access to Anthropic's Fable and Mythos models.
• Anthropic’s Safety Superpower — Anthropic's "safety" narrative strategically aligns with its economic and power imperatives to control frontier AI development and user data.
• Anthropic’s safety warnings may have just backfired — the government has pulled the plug on its most powerful AI — Government intervention in AI model deployment, spurred by developer safety warnings, highlights complex regulatory and commercial tensions.
• Anthropic’s Fable 5 Is Coming Back. The 18-Day Standoff With Trump Is Over. — Government intervention in frontier AI model releases is now a de facto standard due to national security concerns over potential misuse and jailbreaks.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Information and Enterprise Technology News | CIO Dive - Www.ciodive.com.