The Chess.com anti cheat system, as described by the report, is set up to handle blitz and bullet. This makes sense, both because it's most of what is played on

The Chess.com anti cheat system, as described by the report, is set up to handle blitz and bullet. This makes sense, both because it's most of what is played on the site, and because it's so much easier to detect cheating in that environment: A whole lot of engine recommendations would be bananas if a human played them in blitz, and finding major mistakes is very challenging under such time pressure. Evaluating when to play a secondary or tertiary line under pressure, instead of what the engine says, is also challenging. Along with the metadata of thinking time, or whether someone tabbed out of the screen, it's unsurprising that cheating is pretty detectable. On the other hand, in OTB classical chess the difficulty is in bypassing the security controls. Players are much closer to engines, and far fewer moves appear to be completely inhuman. Furthermore, with an accomplice that is also quite good at chess, those lines could just be discarded. This makes it much easier to both play very good moves that are not all that suspicious, and take enough time to "sell" the thinking time. This doesn't mean Hans cheated over the board, just that without very good physical security, any unscrupulous GM could gain 300-400 points virtually undetected, other than from the fact that they are playing much better than they used to. If security is a top priority, at the very least we need to do what the Sinquefield cup did in the last few rounds: No spectators and a 15 minute delay, so that all communication with computers/co-conspirators has to include the player relaying all the moves out before they can get any help.