CaraComp
CaraComp
Forensic-Grade AI Face Recognition for:
Get Started7-day refund guarantee**
Podcast

Hong Kong Police: Facial Recognition Goes Live, No AI Law

Hong Kong Police: Facial Recognition Goes Live, No AI Law

Quick answer

Does Hong Kong have a law on police facial recognition?

No. Hong Kong has no law governing how police use A.I., even as police switch on facial recognition across their camera network. Without a set rule requiring human checks, a software match can be treated as fact. Matches are suggestions that can be wrong, especially on blurry images, so people must verify them.

Hong Kong Police: Facial Recognition Goes Live, No AI Law

0:00-0:00

This episode is based on our article:

Read the full article →

Hong Kong Police: Facial Recognition Goes Live, No AI Law

Full Episode Transcript


Angela Lipps lost her home, her car, and her dog. Police in Fargo had charged her with bank fraud, based on a facial recognition match. Her attorney eventually proved she was in a different state at the time.


If you've ever had your photo taken in public, this

If you've ever had your photo taken in public, this story touches you. Cameras are everywhere, and your face is one of the few things you can't leave at home. Now Hong Kong police say they've switched on facial recognition across their A.I.-enabled camera network. It's part of a four-billion-Hong-Kong-dollar expansion. Police plan to connect sixty thousand cameras to a system called SmartView by twenty twenty-eight. And Hong Kong has no law governing how police use A.I. So when a camera says "that's him," who checks the camera?

Start with Angela Lipps. She's a Tennessee woman who sued a former Fargo detective for ten million dollars. Her lawsuit says Fargo had no policy, no training, and no oversight for facial recognition. A software match pointed at her. Charges followed. The charges were later dismissed, but not before she lost nearly everything. Six months of damage from a guess.

Here's the part that stings. Next door, West Fargo police looked at the same kind of results and made the opposite call. They decided a match alone wasn't enough for charges. They limited the tool to specially trained officers. And they banned arrests based only on a facial recognition match. Same region. Same kind of technology. Different rules, different outcome.

That's the real lesson. The machine didn't decide differently. The people did. For detectives, that's the line between a lead and a verdict. For the rest of us, it means your freedom can depend on which side of a city line you stand.


Trusted by Investigators Worldwide
Run Forensic-Grade Comparisons in Seconds
Detailed facial comparison reports. Results in seconds.
Get Started
7-day refund guarantee**

Consider how well these systems actually work

Now consider how well these systems actually work. A twenty-eighteen M.I.T. study found error rates of nearly thirty-five percent for dark-skinned women. For lighter-skinned men, the error rate was under one percent. That's not a small gap. Researchers also report that police use of facial recognition is associated with a larger gap between Black and white arrest rates. And accuracy drops further on the blurry, grainy images police often have to work with. The sales pitch says the technology is extremely accurate. Real street footage isn't a passport photo.

Then there's the human habit. Research cited in the reporting found some officers call unverified matches a "hundred percent match." A software guess gets spoken of as a fact. The Federation of American Scientists counts at least nine wrongful arrests in the U.S. tied to face recognition misidentification. Nine real people, handcuffed over a probability.

Defenders have a fair answer. They say facial recognition is only a lead, never proof. The New York police department says its matches need review by multiple human analysts. That's a real safeguard. But it's a rule someone has to follow. Fargo had no such rule. Hong Kong hasn't disclosed one. If you're ever the person on the wrong end of a match, a policy that doesn't exist can't protect you.

The danger here isn't that the technology is magic. It's that it's persuasive. A confident answer on a screen feels like certainty, and nobody checks the thing that looks sure.


The Bottom Line

So, in plain terms. Cameras can suggest who someone might be, but they can be wrong, especially on poor images. Police who treat that suggestion as a fact can ruin lives, as Angela Lipps learned. What protects people isn't better software alone. It's rules that make humans check before anyone gets arrested.

Whether you carry a badge or just walk past a camera on your way to work, the question is the same. Who verifies the verifiers? Check the show notes for the full piece.

Ready for forensic-grade facial comparison?

Full forensic reports with detailed similarity scoring. Results in seconds.

Run My First Search