People familiar with investigations estimate tens of thousands of problematic AI incidents involving OpenAI and Anthropic

A reported estimate puts incidents involving OpenAI and Anthropic in the tens of thousands; OpenAI reportedly paused training on its most capable models.