← Retour aux analyses
Cet article n'est pas encore disponible dans votre langue ; la version anglaise est affichée.
IA et Droit
Synthetic data is not a free pass: privacy risk still needs assessment
Training on synthetic data is often seen as privacy-friendly, but if real individuals can be inferred, risk remains.
Synthetic data is often treated as a privacy-friendly alternative, but if it can be reverse-engineered to re-identify real individuals, the privacy risk is not gone.
Regulators increasingly focus on the actual strength of 'de-identification' and 'anonymization', not the label.
Run re-identification risk assessments on synthetic and de-identified data, keep method and test records, and apply stronger protection for highly sensitive cases.