The important question, only important question IMHO, is how they handle positives. Do they go all guns blazing and arrest the person on the spot? Or do they use a restrained approach and first nicely ask the person if they have any ID, etc? That's the important bit.
What if your child falls victim to a false identification, and then given that children are far less likely to have some form of ID on them than adults, they're stuck for much longer?
Do you trust the British police to take good care of your child? Or will they strip-search her and threaten her with arrest like they did with the then-15-year-old Child Q because they decided that she "smelled of weed"?
Do you really want more unnecessary interactions with the police for yourself or those you care about when your "suspicious behaviour" was having an algorithm judge that your face looked like someone else's?
On arrest, you're required to provide your name and address, not proof. For the absolute majority of UK adults, it takes exactly 2 minutes to verify that data against public records - passport, driving licence, council tax, voter registration.
Lying in that situation is a separate criminal offence all of its own.
>satisfy some shit algorithm that misidentified you as some known threat
Matches with a confidence rating of <0.64 are automatically deleted >0.7 is considered reliable enough to present to a human operator, and before any action is taken a serving police officer must verify the match, and upon arrest verify the match against the human.
>What if your child falls victim to a false identification
The age of criminal responsibility is 10, and absent any personal identification parental identification is the standard everywhere.
>15-year-old Child Q
The good old slippery slope fallacy. Both the officers who strip searched that child were fired for gross misconduct. North of 50,000 children are arrested each year and this happened once.
>Do you really want more unnecessary interactions with the police for yourself or those you care about when your "suspicious behaviour" was having an algorithm judge that your face looked like someone else's?
Thing is 12 months on, 1035 arrests, over 700 charges, and that hasn't happened because the point of testing the scheme thoroughly was to stop that from happening.
What proof do you have that it doesn't work.