How Making ChatGPT Safe Takes a Human Toll | WSJ Tech News Briefing