Matej Plavevski comments on The AI Safety Game (UPDATED)