This is not a case of cloaking really, it's an attempt to remove temporal data from a static cache-- the only other ways to do that would be to remove the temporal data for non-registered clients, or to include a no-cache directive in robots.txt. Both are unacceptable since the former means AM can't see chat, and the latter because it makes all of PM non-cached. Since the primary "offender" is Google, I'd simply look at their UA string and give them a different page based on that. But I'm not an EE hacker, so I simply offer this as a "nice to have" to the development team.

I have to wonder how well cloaking-detection even works without human intervention... you can't simply compare HTML from one GET to the next, the site could be using UA to send tuned HTML, or could have a random feature, or any number of other things resulting in slightly dissimilar HTML results. As such, it would almost have to undergo human review, or some similarity testing that PM, with or without Chatter, would probably pass.