A side-channel vulnerability in one of the world's leading post-quantum encryption standards has been demonstrated by researchers at SEALSQ, who presented their findings in a peer-reviewed study at LightSec 2026. The work targets ML-KEM, the lattice-based key encapsulation mechanism selected by the US National Institute of Standards and Technology as a core building block for encryption designed to withstand quantum computers. The study matters because ML-KEM is already being built into hardware and software meant to protect data for decades to come.
Why ML-KEM Is Under Scrutiny
ML-KEM, derived from the CRYSTALS-Kyber algorithm, relies on the mathematical difficulty of solving certain lattice problems, a class of math believed to resist attacks from both classical and quantum computers. It has become a central candidate for replacing RSA and elliptic-curve cryptography, which quantum algorithms could eventually break. But an algorithm's theoretical strength does not guarantee the safety of its implementation. Side-channel attacks do not try to break the underlying mathematics at all. Instead, they observe physical signals - power consumption, electromagnetic emissions, timing variations - produced by a device while it performs cryptographic operations, and use those signals to infer secret information.
Key generation is a particularly sensitive phase, because it is when the private key material is first created and handled in memory or on a chip. If an attacker with physical or close proximity access can capture enough side-channel data during this process, they may reconstruct fragments of the key without ever touching the mathematics that is supposed to protect it.
A Hybrid Attack Method
What distinguishes the SEALSQ research is the combination of two established techniques into a single, more efficient attack. The first is profiling, a method in which an attacker builds a detailed statistical model of a device's physical behavior, typically using a copy of the target hardware under controlled conditions. The second is lattice reduction, a mathematical technique used to simplify complex lattice problems into more solvable forms. By pairing a physical profiling stage with a mathematical reduction stage, the researchers showed it is possible to extract sensitive key data with greater efficiency than either method achieves alone. This hybrid approach narrows the gap between theoretical side-channel risk and a practical, repeatable attack - precisely the kind of result that hardware manufacturers and standards bodies need to see before vulnerabilities are exploited in the field rather than in a laboratory.
Countermeasures and Industry Implications
Alongside the attack, SEALSQ proposed practical defenses intended to harden ML-KEM implementations against this class of threat. Side-channel countermeasures typically involve techniques such as masking, which randomizes intermediate computations so that physical signals no longer correlate cleanly with secret values, and constant-time execution, which prevents timing differences from leaking information. The company has positioned this research as part of the engineering work behind its Quantum Shield QS7001, a secure microcontroller platform built for post-quantum cryptographic workloads.
The broader lesson extends well beyond one company or one chip. As governments and industries migrate toward quantum-resistant standards, the security of those standards will depend not only on sound mathematics but on disciplined, verifiable hardware engineering. Side-channel research of this kind is a normal and necessary part of that maturation process, exposing weaknesses before attackers do and pushing implementers toward more rigorous safeguards. For organizations planning long-term data protection strategies, the finding is a reminder that quantum resistance on paper is not the same as quantum resistance in a deployed device.