Anthropic has begun rolling out the latest versions of its Claude AI assistants, with Claude Mythos 5.2 now available to a limited set of partner organizations and Claude Fable 5.2 released to the broader user base. Both updates are described as delivering enhancements to coding assistance and logical reasoning capabilities.
According to the company’s release notes, Claude Mythos 5.2 shows a marked improvement in its ability to locate software bugs. The model is reported to be more effective at spotting coding errors and potential security flaws, offering developers a sharper tool for code review and vulnerability identification.
In contrast, Claude Fable 5.2 is said to block fewer security‑related tasks than its predecessor. The change means the assistant is less likely to refuse prompts that involve security testing or analysis, potentially expanding its usefulness for users who need more permissive access to security‑focused workflows.
Anthropic highlighted that these adjustments are part of an ongoing effort to balance safety controls with functional flexibility. The company indicated it will continue to monitor how the new behavior impacts both developer productivity and the broader security landscape.
Founded by former OpenAI researchers, Anthropic’s Claude series has become a prominent competitor in the generative‑AI market. The latest 5.2 releases reflect the firm’s strategy of iterating quickly on model performance while fine‑tuning the trade‑offs between security safeguards and practical utility for its users.
<small>Source: Bleeping Computer — read the original story there.</small>