That's not entirely as accurate as phrased. More accurately, researchers have discovered side-channel attacks that make an old attack effective again. The act

That's not entirely as accurate as phrased. More accurately, researchers have discovered side-channel attacks that make an old attack effective again. The actual "unmasking" technique here is actually not new at all. What is new is the ML-based side-channel that leaks whether images or other protected resources are successfully loaded.