[DoctorOetker]

I now recall that the claim was not that the image sensor itself fluoresced or raman scattered or whatever in a characteristic way, but that it was the IR blocking filter before the sensor...

this immediately makes the setup less usefull for surveillance cameras as they will typically not be optimized for color fidelity, but for light sensitivity, so without IR filter...

Nevertheless I would still be interested in more details or the exact mechanism or references...

[vertexFarm]

That's a good point. The IR filter should light up like a beacon with most red LEDs that include a bunch of IR, such as the ones you see in a ring around most night-vision security cameras. I don't know if that would be detectable with just the naked eye and a filter, though. You'd have to have another camera capable of seeing the IR.

However it wouldn't detect those same night-vision cameras, since they are most often just standard webcam guts with that IR filter omitted as you say.. I really think this thing is just what it says--a lens detector, which catches specular glints off the glass lens and doesn't have much to do with the actual CCD or CMOS or whatever sensor is underneath. It could definitely be making use of the IR filter on most cameras to make those shine particularly bright.

[DoctorOetker]

yes I also for no second believe this cheapo led ring with filter is the hypothetical camera detector used in cinemas, but that instead it is based on reflections. A voyeur-run room could simply use lots of glass beads in an artistic way to decorate the room like glueed to the walls etc.

I'm not sure how the cinema version works though, or might the IR filter be a dielectric high-pass filter (similar to how anti-reflective coatings work), passing red but reflecting near IR?

Optically in the near IR that would be equivalent to a mirror between sensor and lense, for simplicity lets pretend the filter is nearly in the same plane as the sensor in the focal plane, then parallel rays focused at infinity would converge to a point on the mirror, which means the light should retroreflect, but the intensity of retroreflection should be highest at the optical axis, and quickly fall off away rom the axis... that makes for 4 DOF to "scan" direction to aim the beam of light, and 2-dimensional position with respect to aperture... so you need to be either lucky or have a very good idea of typical orientation and position of the camera, which would also explain the usage in cinema's you know the orientation of the camera so now only 2d of freedomm left...