Aside from all the moral and ethical questions this raises from the usage of faceless AI, really this happens time and again, I suspect so Google can employ less human reviewers and so they can claim "not our fault". I think I've commented on a couple of these posts by now every time with the same advice.

But anyhow, people should be well aware cloud providers do use AI systems and they will make mistakes, so you should:

1: have multiple redundant backups, seriously this is a no-brainer, AI or no even Google can lose data if a storage device dies. Depending on how important the data is you might consider as many backups as you can manage.

2: encrypt, encrypt, encrypt. If you store even so much as a text file unencrypted you're making a mistake. Use archives, double compressed, use the myraid of tools out there to ensure your files can't be read by anyone but you and who you decide to give the key to.

Is this more work? Absolutely, but at the end of the day regulations mean Google need to scan content, and they will have a false positive match, and then it goes on HN and magically Google restore the data.

I hope they get the data back but sometimes being burnt is actually a lesson in remembering best practice, why oh why your only copy of family photos are on one user hostile data-mining platform is beyond me, but it's a terrible idea.