Jordan Taylor comments on Sabotage Evaluations for Frontier Models