An IRB would only care about whether the deception perpetrated on the subject is justifiable. They wouldn't care one iota how the light was disabled--the risk to the subject is the same regardless of whether the light is disabled by hardware or software or not used. For example a BYOD type study might be plausible because in some circumstances subjects may be more comfortable/candid using their own devices. Who knows.

The issues for the IRB would only be the fact that the subjects are not consented to covert surveillance and whether the nature of the information collected presents a risk to the subject if accidentally disclosed.