Tuesday 28 July 2026 17:35:08 GMT+02:00
Netcrook
#anonimizzazione dati
The AI Privacy Trap: Why “Anonymous” Data Can Still Come Back to Bite
In AI pipelines, de-identification is not a magic switch - the real question is whether re-identification risk remains defensibly low after reuse, sharing, and model training.
When Data Opens Up, Privacy Becomes the Real Attack Surface
The DMA’s push for search-query sharing under Article 6(11) puts anonymization at the center of a difficult question: what level of protection is enough to keep re-identification risk under control?