Anthropic opens public access to Fable after warning of AI risks

Anthropic released Fable to the public on Tuesday after previously keeping the model restricted because of concerns that its capabilities could be dangerous. The company said the version now available includes safeguards and usage limits, while acknowledging that a model this capable carries risks. Anthropic also expanded access to Claude Mythos 5 for about 150 organizations that had previewed the system, including a version without cybersecurity or biology limits for certain approved uses. The company said testers using Mythos have uncovered more than 10,000 critical security flaws. Finance Minister François-Philippe Champagne had earlier said concern over Mythos was justified because of the "unknown, unknown," and US government agencies have also been testing the tool.