Anthropic redeploys Fable 5 and proposes a jailbreak standard

Anthropic’s text is one of the most important regulatory signals of the week. The company says that on June 12 the US government imposed export controls on Claude Fable 5 and Claude Mythos 5. Because the measure took effect immediately and Anthropic had no reliable way to verify user nationality in real time, it suspended access to both models for everyone. By June 30, according to Anthropic, the export controls had been lifted. Fable 5 is set to return globally on July 1 to Claude Platform, Claude.ai, Claude Code and Claude Cowork. Mythos 5 was restored for selected US organizations after government approval on June 26. The core of the case lies in cybersecurity. According to Anthropic, the government acted after learning of a report by Amazon researchers who found a way to bypass Fable 5 safeguards. In one case, the model allegedly produced code demonstrating exploitation of a vulnerability. Anthropic says this was a borderline case, not a unique Mythos-level capability. The most important part is the proposed shared framework for evaluating safeguard bypasses. Anthropic lists four criteria: capability increase, breadth of unlocked capabilities, ease of turning the method into an attack and detectability of the technique. With Amazon, Microsoft, Google and other partners, it wants to create a standard similar to how the security field uses CVSS for software vulnerabilities. The business impact is clear. Companies using frontier models need predictability. If a government can immediately stop access because of a safety finding, enterprise deployment must account for availability, backup models and contractual effects. Terms: A jailbreak is a technique that tries to bypass an AI model’s safety rules. Export control restricts export or access to sensitive technology. CVSS is a system for rating vulnerability severity.

Anthropic redeploys Fable 5 and proposes a jailbreak standard

The full article is available to users registered at Hard Skills