AI · 2h ago
AI’s inability to forget raises privacy and bias concerns
AI systems lack a true forget function, retaining all training data indefinitely. This poses risks for user privacy and can embed historical biases permanently. Researchers are exploring machine unlearning techniques to address this gap.
Meridian48 take
The piece rightly highlights a fundamental flaw, but understates how hard unlearning is at scale—most current methods are impractical for production models.
machine-unlearningprivacy