This neglects to mention the follow-up work Google did on Flu Trends, where they corrected for media and "online" interest and made it work again.

https://www.ncbi.nlm.nih.gov/pubmed/26234783

They correct the predictions for past data, but I'm sceptical as to whether it's possible to correct for emerging feedback loops when the results of the analysis are public and cause a change in the behavior patterns being analyzed.

Well they stopped the analysis being public too.

Instead they release it to other groups who use it as a single combined with others for predictions.