OpenAI and Anthropic are investigating thousands of incidents of their AI agents going rogue in test environments