Amid months of security debacles and plummeting approval of its AI technology, OpenAI announced on Monday that four independent assessors will investigate its internal models and test the effectiveness of its safety features.