Here's my reasoning:
Setup 1: Take a traditional double slit experiment and turn on the photon source. Interference fringes should appear on a detector screen placed opposite the photon emitter, right?
Setup 2: The often cited "which-way" variation of the double slit experiment, in which extremely sensitive photon detectors are placed in the slits. These photon detectors are designed to allow the photons to pass, but to measure their passage. The interference fringes on the detector screen should no longer be present and the detectors placed at the slits should have a record of the passage of the individual photons.
Setup 3: A "which-way" variation with the photon emitter turned down so that it emits only 1 photon per second. The result, at least on the detection screen, of setup 3 should be the same as setup 2, just really, really dim. However, now that the rate of photons' emission and detection is drastically reduced and carefully controlled, there is an additional piece of information available. That information is the "time stamp" of each photon emission/detection event. Using that additional time-stamp information, correlations between the two slits can be recorded for each specific photon. For example, the measurements from slit A are recored by Alice and the measurements from slit B are recorded by Bob. Now, if I'm not mistaken, Alice and Bob should have a perfect one-time pad which can be used for unbreakable encryption.
I've heard of quantum entanglement enabling secure one-time pad distribution, but the photons in a double slit experiment are not necessarily entangled in the traditional sense. And yet, the combination of the time stamp information and the photon's necessarily complementary passage through only one of the two slits available would seem to imply that a secure one-time pad can be distributed (with proper planning and implementation), without the need for a shared source of entangled photons.