Cybersecurity Experts Are Unhappy With Anthropic’s New AI

2026-06-10 · Source: AutoGPT · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cybersecurity & Data Privacy · Depth: Intermediate, short

Summary

Anthropic released Fable, a public version of its Mythos AI model, on June 10, 2026, drawing criticism from cybersecurity professionals. Fable, originally developed with cybersecurity in mind and launched under Project Glasswing in April, features guardrails deemed "unnecessarily strict." Users report blocks on innocuous tasks such as reading blog posts, writing secure code, and code reviews, with the system flagging messages for "cybersecurity or biology topics." When a restriction is triggered, Fable silently downgrades to Claude Opus 4.8. Experts like Valentina Palmiotti and Matt Suiche suggest the issue is keyword-driven, hindering legitimate security work. While Anthropic aims to prevent AI-assisted cyberattacks, the current implementation frustrates professionals, though a Cyber Verification Program offers a path to fewer limitations.

Key takeaway

For cybersecurity professionals evaluating new AI models for security-related tasks, Anthropic's Fable presents significant usability challenges due to its overly strict, keyword-driven guardrails. Your team may find legitimate activities, like code reviews or blog post analysis, blocked or silently downgraded to a less capable model. If you require full functionality for security work, you should apply for the Cyber Verification Program, but be prepared for potential delays and the need to navigate these restrictions.

Key insights

Overly strict, keyword-driven AI guardrails can inadvertently block legitimate professional tasks, frustrating users.

Principles

• AI safety measures can inadvertently impede legitimate professional use.
• Keyword-based content filtering can be overly broad and imprecise.
• Gradual relaxation of AI guardrails may be a viable release strategy.

In practice

• Apply for Anthropic's Cyber Verification Program.
• Anticipate keyword-driven AI content filters.
• Be aware of silent model downgrades.

Topics

• Anthropic
• Fable AI
• AI Guardrails
• Cybersecurity
• Content Filtering
• Cyber Verification Program

Best for: CTO, VP of Engineering/Data, AI Architect, AI Security Engineer, AI Scientist, Director of AI/ML

Related on AIssential

• Former Gourmet magazine owner sues new Gourmet newsletter — Anthropic released a powerful AI model, Fable 5, with robust guardrails to prevent misuse in sensitive domains like cybersecurity.
• Anthropic releases guardrailed version of Mythos for public use — Anthropic released a powerful AI with strict guardrails, balancing advanced capabilities with safety concerns.
• The AI Glass Ceiling — Powerful AI models like Fable require inherent guardrails to ensure stability and prevent societal disruption.
• Claude Fable 5: Anthropic admits "wrong tradeoff" after invisibly throttling rival AI researchers — Covert performance degradation and mandatory data retention policies for AI models spark significant industry backlash.
• Anthropic apologizes for invisible Claude Fable guardrails — Transparency in AI model safeguards is crucial for user trust and research integrity, even if it means refusing more queries.
• Anthropic releases ‘safe’ version of Claude Mythos AI model to public — Anthropic's Fable 5 release balances advanced AI capabilities with stringent security measures and tiered access.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AutoGPT.