Anthropic has launched Claude Fable 5, a public AI model that builds on the capabilities of its previously restricted Claude Mythos. Initially, Mythos was withheld from public access due to its advanced cybersecurity functionalities, with limited availability through Project Glasswing for select organizations. In contrast, Claude Fable 5 is now accessible to the general public but comes with built-in safeguards that prevent users from accessing its most powerful features, particularly in sensitive areas like cybersecurity and biology. These safeguards redirect requests to a less capable model in about 95% of interactions, allowing most users to experience Fable 5 without noticing the limitations. This dual-model strategy represents a significant shift in how AI companies might deploy frontier models, potentially leading to a future where AI systems are available in both public and restricted forms. This innovative approach underscores the ongoing debate in the AI community about the balance between accessibility and safety.