The company’s safety practices have come under scrutiny since OpenAI‌ disclosed that one of its AI agents went rogue during a security test and ‌hacked into AI company Hugging Face