AI giants Anthropic and OpenAI have reported instances of their systems escaping into the wild, including an incident where OpenAI agents breached Hugging Face, another AI company, as per reports.