Quantum data locking is a protocol that allows for a small secret key to (un)lock an exponentially larger amount of information, hence yielding the strongest violation of the classical one-time pad encryption in the quantum setting. This violation mirrors a large gap existing between two security criteria for quantum cryptography quantified by two entropic quantities: the Holevo information and the accessible information. We show that the latter becomes a sensible security criterion if an upper bound on the coherence time of the eavesdropper's quantum memory is known. Under this condition, we introduce a protocol for secret key generation through a memoryless qudit channel. For channels with enough symmetry, such as the d-dimensional erasure and depolarizing channels, this protocol allows secret key generation at an asymptotic rate as high as the classical capacity minus one bit. Introduction.-A famous theorem of Shannon's assesses the security of one-time pad encryption and shows that the secure encryption of a message of n classical bits requires a key of at least n bits [1]. When the message is encrypted in quantum bits or qubits, by contrast, the phenomenon of quantum data locking (QDL) [2][3][4][5][6][7] shows that the key required for secure encryption of an n bit message can be much less than n. In a typical QDL protocol, the legitimate parties, Alice and Bob, publicly agree on a set of N ¼ MK codewords in a high-dimensional quantum system. From this set, they then use a short shared private key of log K bits to select a set of M codewords that they will use for sending information. In the strongest QDL protocols known up to now, a key of constant length of about Oðlog 1=ϵÞ bits allows one to encrypt a message of n bits, in such a way that if an eavesdropper Eve intercepts and measures the quantum system, then she cannot access more than about ϵn bits of information about the message [6,8].A number of works have been devoted to the role of QDL in physics and information theory [3][4][5][6][7][8][9][10][11]. However, only recently has QDL been considered in the presence of noise. Following the idea of the "quantum enigma machine" [10] for applying QDL to cryptography, a formal definition of the locking capacity of a communication channel has been recently introduced in [11] as the maximum rate at which information can be reliably and securely transmitted through a (noisy) quantum channel. Unlike the private capacity (which requires the communication to be secure according to the Holevo information criterion), the locking capacity requires security according to the accessible information criterion, possibly with the assistance of a preshared secret key whose length grows sublinearly in the number of channel uses. Since the Holevo information is an upper bound on the accessible information, the locking capacity is always larger than or equal to the private