Possible defenses against the attack vector of facial recognition via unwanted photographs.
The advance of facial recognition technology has promoted unwarranted photographs from being a nuisance to being a potentially substantial threat, for recent examples see the cases of Clearwater AI and PimEyes. Although there are many angles to this issue, right now my interest is in technical offlince solutions for avoiding having one's picture taken without consent.
What I found
It seems like the best defenses rely on infrared light.
- This answer, proposes the use of LEDs to hide your face from cameras via overexposure, although there are some drawbacks like the need for a constant power source.
- A vendor of IR-reflecting glasses promises blocking "95% of the near infrared radiation spectrum". They also sell a frame that produces sort of a halo effect, covering much more than just the eyes. The main drawback here is that those glasses are not super cheap (84$ to 164$), and although privacy always comes at some cost, this could be too high for many people.
- According to this answer, certain types of welding glasses should block the relevant spectrum. However, I don't know to what degree they block it and if this would be enough to hide your eyes from cameras.
There are other methods as well, like obscuring your face with a scarf, balaclava, or even the more refined URME mask, but all of these look suspicious to some degree (even though the URME mask is supposed to look innocuous apart from the missing facial reactions).
What is the best way to avoid having an identifiable picture taken, with the restraint that it should be a) accessible and b) relatively inconspicuous?
Since several comments mentioned some kind of masking: Simply obscuring your face (by wearing a mask, scarf or even make up) no longer seems like a viable option, since Chinese media already report facial recognition technology that works even for obscured faces