A question I've always wondered -- If Google, or an employee at Google, used ML (or just classic techniques) to analyze the enormous, enormous troves of trojan horse data they have on the employees of virtually every organization across the globe, would it be illegal if they traded on that data? They could surely accurately call virtually any trend before anyone. They could see when morale at a firm drops, when hours…
As such, they do not have "enormous troves of trojan horse data" as they are not allowed (by their own terms) to use it for anything unrelated to the company itself.