Anonymized data re-identified with high accuracy
Categories: technology, governance
Tags: privacy, ai, datascience
What? Just as Googling a quoted string of words can return a unique hit, in 2015 MIT researchers showed that supposedly anonymized strings of purchases can be used to effectively identify individual shoppers. The researchers found that from only four shopping receipts containing location information associated with a purchase it was possible to re-identify an individual with 90-percent accuracy.
So what? In the future, deep learning artificial intelligence software may be used to further improve metadata-based re-identification of anonymized data, challenging the very nature of this kind of privacy protection. This will render the current legal concept of personally identifiable information, on which privacy laws depend, inadequate for metadata sets.
Source: The Register - A docket, tweet and selfie can reveal your identity, boffins find