Unconditional security — encryption that no computational advance can break — has been the province of quantum key distribution. QKD exploits the no-cloning theorem: intercepting a quantum channel necessarily disturbs it. But QKD requires optical fiber or line-of-sight, limiting range. The security guarantee lives in physics, specifically in quantum mechanics.
This paper moves it to chemistry.
Two parties share physically duplicated pools of synthetic DNA molecules, each containing random index-payload pairs. Tokyo and Paris each receive identical molecular samples. Each independently sequences their sample, digitizes the results, and extracts binary masks. The result: approximately 400 Mb of shared secret with 2^{-128} decryption failure probability. A one-time pad, generated from molecular biology.
The security guarantee rests on the same principle as QKD but through a different physical realization. Tampering is detectable because molecular copy-number statistics have characteristic distributions. Intercepting the DNA sample and attempting to copy it introduces statistical anomalies — deviations in expected molecule counts — that reveal the interception. You cannot copy the molecules without disturbing the population statistics, just as you cannot copy a photon without disturbing its state.
The advantage over QKD is range. DNA can be shipped anywhere physical matter can travel. There is no distance limitation, no fiber requirement, no atmospheric absorption. The shared secret is a vial, not a laser pulse.
What makes the result structurally interesting is the generalization it implies. Information-theoretic security requires a physical channel where copying is detectable. Quantum mechanics provides one such channel. Molecular statistics provide another. The security lives not in any particular physics but in the thermodynamics of replication — the same principle that makes genetic fidelity possible.