<p>After OpenAI's models broke into Hugging Face, Anthropic checked its own history and found three similar incidents</p>