Anthropic has restored access to its Fable 5 model globally after "a series of productive conversations with the US government," redeploying it with "a new set of classifiers to target and block more cybersecurity tasks." The company confirmed that Amazon researchers identified "a method of bypassing Fable 5’s safeguards," and subsequent testing "confirmed that many less capable models—including Claude Opus 4.8, GPT-5.5, and Kimi K2.7—could identify the same vulnerabilities," though "the reported technique did not expose any unique Mythos-level cyber capabilities." Anthropic will work with Amazon, Microsoft, Google, and other Glasswing partners on "a consensus framework for assessing the severity of AI jailbreaks." Fable 5 will be available on the Claude Platform, Claude.ai, Claude Code, and Claude Cowork, included for up to 50% of weekly usage limits through July 7 on certain plans. Access to Mythos 5 has also been restored "for a set of US organizations" that operate and defend critical infrastructure, following a government order that had cut off access earlier this month due to concerns about potential misuse and alleged ties to China involving South Korea's SK Telecom.