Phys. Rev. Lett. (in the press); preprint at http://arxiv.org/abs/1106.3224

Quantum mechanics permits stronger correlations between remote measurements than are allowed classically. A routine procedure to test whether a pair of particles is engaged in such a non-classical liaison is through the violation of a 'Bell inequality'. The procedure reveals the presence of quantum correlations independent of the physical device used based on purely statistical criteria, and therefore in an unambiguous manner.

Or does it? A note of caution comes from Ilja Gerhardt and colleagues, who have demonstrated that in a typical experimental setting the violation of Bell's inequalities can be faked using purely classical light. Their 'faked-state generators' can cause avalanche photon detectors — standard components in such experiments — to produce any desired measurement outcome by sending bright light pulses instead of single photons.

Even when the legitimate users have full knowledge of their measurement devices and observe no unusual behaviour, an adversary can fool them into seeing quantum correlations where there are none. Similar to the conclusion reached in related work (New J. Phys. 13, 063031; 2011), Gerhardt et al. warn that the assumptions underlying Bell tests, in particular regarding to the so-called detection loophole, have to be meticulously observed.