Published • loading... • Updated
Anthropic rolls out public version of Mythos without cybersecurity capability
The model blocks high-risk requests and falls back to Claude Opus 4.8, while Anthropic says early testing found no universal jailbreaks in over 1,000 hours.
- Anthropic officially launched "Claude Fable 5" on Tuesday, commercially releasing a public, heavily safeguarded version of its highly powerful "Mythos" artificial intelligence architecture.
- The debut follows a highly restrictive two-month private rollout of the original "Mythos Preview" model in April, which alarmed Wall Street, central banks, and global governments due to its unprecedented ability to automatically find and exploit critical software vulnerabilities.
- Fable 5 represents a massive leap in raw analytical power, marking the first public AI model to break 90% on the core Hex benchmark for highly complex, multi-run, and long-horizon analytical tasks.
- To mitigate severe misuse risks, the model features aggressive routing guardrails; any prompt asking about high-risk topics like cybersecurity or biology is automatically downgraded to the less permissive Claude Opus 4.8, triggering a fallback in roughly 5% of user sessions.
- Simultaneously, a fully unrestricted "Claude Mythos 5" was deployed via the defensive cybersecurity initiative "Project Glasswing," keeping the raw, un-guardrailed model strictly limited to a small group of infrastructure providers and U.S. government cyberdefenders.
Insights by Ground AI
Podcasts & Opinions
83 Articles
83 Articles
Anthropic Launches Claude Fable 5 and Claude Mythos 5, the Most Powerful AI Models He Has Ever Built
Anthropic has just released the two most powerful artificial intelligence models in its history, and it all points to the industry never going to be the same again.
Anthropic releases a less-powerful version of its most advanced model
Anthropic on Tuesday unveiled a new “Mythos-class” model for general use, as concerns around the release of advanced artificial intelligence tools with powerful hacking capabilities deepen. The new model, Fable 5, performs at a level similar to Anthropic’s most advanced Claude Mythos model, though the company says it has built-in safeguards to prevent users from prompting it to carry out a cyberattack or to detail how to create a bioweapon. Its …
Coverage Details
Total News Sources83
Leaning Left22Leaning Right10Center19Last UpdatedBias Distribution43% Left
Bias Distribution
- 43% of the sources lean Left
43% Left
L 43%
C 37%
R 20%
Factuality
To view factuality data please Upgrade to Premium
