I fail to see the connection. If Amazon began to analyze recordings to try and find real or even misconstrued abuse (presumably only after legislation required them to - I can't see a business case), they'd reasonably need to adjust what their algo flagged as a match. I'd be curious if Echo (or gHome, or Siri) do any sentiment analysis right now, they don't seem to (or cursing when they botch your command would at least produce more contrition).