#identity documents in datasets

A new study reveals that millions of personal documents and identifiable faces are present in one of the largest AI training datasets, highlighting significant privacy risks in AI development.