Comment by xiphias2

Models at this point know about chain-of-thought monitoring so they already know they need to hide the cheating, it's just a matter of time they start doing it