A claim is circulating online that Anthropic's most powerful and "too dangerous to release" AI model — codenamed Claude Mythos — has quietly appeared in the Google Cloud console with its "Preview" label stripped, supposedly repeating the exact rollout choreography that preceded the public launch of Claude Opus 4.7.