> We hear a lot about the potential for sabotage, it’s extremely rare in practice, and we have built some safeguards in place to prevent sabotage from succeedin

> We hear a lot about the potential for sabotage, it’s extremely rare in practice, and we have built some safeguards in place to prevent sabotage from succeeding This wouldn't ring so hollow if Google didn't have such a reputation for blindly automating fraud detection. I seriously doubt it is actually possible to differentiate the two. The fraudulent activity that is visible to Google is the same regardless of the intention of the client that paid for it, be it a beneficiary or an adversary. Any signal that is deemed fraud by Google can eventually be simulated by an adversary.