Roughly a dozen OpenAI employees knew about concerning chatbot interactions linked to a Canadian school shooter, but the company chose not to notify law enforcement, according to a new report.