- Anthropic has restored public access to Claude Fable 5 and Mythos 5 after the US government lifted export controls following a security review.
- Claude Fable 5 now includes stronger AI cybersecurity classifiers designed to detect and block a wider range of cybersecurity-related tasks.
- The US government approved the models after working with Anthropic to improve AI safety while supporting America’s AI leadership.
A new set of classifiers intended to recognize and prevent a wider range of cybersecurity-related tasks will be used to redeploy Claude Fable 5. Claude Fable 5 and Mythos 5, two of Anthropic’s most potent AI models, will now again be available to the public, weeks after the US government ordered their removal.
Anthropic, an American artificial intelligence (AI) company known for developing advanced and safety-focused AI models, had restricted public access to its two latest models, Claude Fable 5 and Mythos 5, since June 12. The move came after the US government imposed export controls following a report that researchers had bypassed Fable 5’s safety safeguards.
Anthropic immediately suspended public access to both models. The company announced that the US government lifted the restrictions on Wednesday, allowing the models to return with enhanced security measures.
Claude Fable 5 will be available again globally tomorrow.
After a series of productive conversations with the US government, we’re redeploying the model with a new set of classifiers to target and block more cybersecurity tasks. In the near term, some routine tasks like coding…
— Anthropic (@AnthropicAI) July 1, 2026
“We’re redeploying the model with a new set of classifiers to target and block more cybersecurity tasks after a series of fruitful conversations with the US government,” Anthropic stated.
US & Anthropic Strengthen AI Cybersecurity Efforts
Experts and technologists said the models’ suspension created a risky precedent and raised questions about state control over cutting-edge AI technology. The White House’s worries about a possible national cybersecurity danger if these potent models were jailbroken and used maliciously were also brought to light by the export restrictions.
“Over the past two weeks, we have worked closely with Anthropic to analyze and approve Fable 5 to ensure alignment across the US Government and to strengthen America’s leadership in AI,” US Secretary of Commerce Howard Lutnick stated on Wednesday.
The government’s top goal is still to “get the best AI tech deployed as quickly and safely as possible,” according to White House Chief of Staff Susie Wiles.
The limitations were implemented after the government learned of a study in which Amazon engineers discovered a way to get over Fable 5’s security measures, causing the model to find many software flaws.
AI Jailbreak Prompts Tougher Claude Fable 5 Protections
Anthropic said that this risk wasn’t specific to Fable 5 because weaker models might also find the same flaws and generate the same exploit
As part of Project Glasswing, a partnership established in April to protect against AI cybersecurity threats, Anthropic has also started developing a consensus framework for “assessing the severity of AI jailbreaks” with Amazon, Microsoft, Google, and other partners.
The corporation is increasing its cooperation with the US government on safety and testing of AI models. It said, “This will include information sharing on jailbreaks and misuse, pre-release access to models and safeguards for evaluation, and dedicated resources for joint research.”
A well-known AI researcher posted screenshots demonstrating how he got over the model’s security measures and claimed to have jailbroken Fable 5 within 48 hours of its June launch, prior to the government limitations.
Stay informed with the latest trends in Web3, blockchain innovation, and cybersecurity updates at 3verseTV
You need to login in order to Like
Leave a comment