it could easily apply the same to personal data to be flagged
Please do enlighten us how it could easily apply algorithms to categorize data to distinguish between personal, protected data, and data of public records that belong to someone else. Just for shits and grins, please create an algorithm that would distinguish between the Washington Post article and the original bankruptcy article.
It's perfectly possible to have both- no one is expecting perfection, but ultimately just because Google may never get it perfectly right doesn't mean they should be freed from the law altogether.
Wow. So that means that now laws that cannot be followed every time are a good idea? In the case of Google, it means a perpetual fine that cannot be escaped, is completely arbitrary, and applies only to Google.
Everything you posted so far is a damning indictment of exactly why this law is terrible: it's not possible to fully comply, it's arbitrary, it's open to abuse from all sides, and its target is also completely arbitrary.
Technically, you are accurate in your description of why Google needs to follow the law as it is written. However, the discussion we're having is about whether the law should exist in the first law. On that, you're digging your own hole.