Years ago, I worked on a project that involve capturing images filmed with an ophthalmoscope and mosaicing them into a complete image of the back of the eye. Part of that involved finding out how bright a light we could get away with beaming into peoples eyeballs. I don't remember specifics, other than being surprised at how much you could get away with.

A quick search showed that there are standards that cover this and include infrared light (see http://www.fda.gov/MedicalDevices/DeviceRegulationandGuidanc... for more information).