"Possibility for abuse" seems like the right reason here. Does the benefiting of reducing a specific possibility of abuse outweigh the cost of an intervention? And here in particular, is there much cost to the intervention other than just shifting the money distribution from a zero-sum advertising arms race from one player to several?

I frequently see calls to not intervene if there's not bulletproof evidence of existing abuse, but why wait? Would you want Google to own a bunch of nuclear missiles just because they might not have misused them yet?