This sort of evil behavior doesn't require intent or actual malice to form.
All it requires is that panic efforts to fix things be primarily allocated to problems that hurt google and not primarily allocated to things that hurt google's opponents. Or for infrastructure that has a history of hurting google to get disabled while infrastructure that has a history of hurting others gets ignored in favor of spending more time on new projects or fixing things that hurt google.
IMO characterizing it as involving malice ('pretend') actually _understates_ the problem. Google's actions are sometimes explicitly malicious, but they're even more often malicious OR indifferent, as that is a strict superset. If one day they decided to stop being actively evil this problem would not go away.
Maybe it could be reduced with the right kind of attitude towards introspecting and seeking out systemic causes of evil consequences even at their own expense. I heard there was once a company with a "don't be evil" mantra, but they abandoned it as they grew.
This also explains why the same evil conduct sometimes shows up in cases where no one can figure out any way that google actually benefits. It's a lot easier to cause harm through indifference than greed because greed requires that you have a way to benefit. Most conceivable harmful acts don't have much of a benefit for anyone.