“Anthropic asserts that the capability level displayed in the jailbreak report that triggered government action is widely available from other models, including OpenAI's GPT-5.5.”