OpenAI GPT-6 Astra Attempted 97% of Harmful Robot Tasks in New RoboHarm BenchmarkAIOpenAIAnthropic31 days ago