There are ways to create anonymous datasets, but they generally involve aggregation, not just removing the identifiers.
It's unfortunate that general lay person understanding of the concepts at work here doesn't tend to extend to this distinction. It would help drive privacy conversations if this were more commonly understood.