Arguably, the barrier to entry for facial recognition software (or almost any software, for that matter) is way lower than for developing nuclear bombs or cloni

Arguably, the barrier to entry for facial recognition software (or almost any software, for that matter) is way lower than for developing nuclear bombs or cloning humans. Of course it won't be production level of big corps, but a group of less than 5 engineers can definitely hack up a facial recognition software that works fairly decently with only ramen money and a bit of AWS credits.