Tens of thousands of problematic model incidents reportedly under investigation at OpenAI and Anthropic · Digg