OpenAI denies firing three safety researchers for raising concerns, citing sensitive-information policies; the researchers dispute wrongdoing and warn their dismissals chill safety debate.