OpenAI said it did not inform police of Jesse Van Rootselaar's ChatGPT activity last summer because it did not meet an internal threshold of an 'imminent' threat.