5

According to this recent NBC story, New facial recognition tech catches first impostor at D.C. airport, a man arriving on a flight to the US presented a passport to the customs officer, but using the new facial comparison biometric system, the officer determined the traveler did not match the passport he presented.

The 26-year-old man arrived Wednesday on a flight from Sao Paulo, Brazil, and presented a French passport to the customs officer, according to the U.S. Customs and Border Protection (CBP). Using the new facial comparison biometric system, the officer determined the unidentified traveler did not match the passport he presented.

I am skeptical that the person presented in this story was really primarily caught by the facial comparison technology. If you are a European citizen arriving at the US border, you first have to hand over your passport to the custom officer, who scans it with the computer and then takes a picture of you and a fingerprint scan of both of your index fingers. After this, you briefly have to explain why you want to enter the US.

French (as well as most other European) passports have had for many years the fingerprints of the person digitally stored in a machine readable chip. According to this wikipedia article on biometric passports

To store biometric data on the contactless chip [...], it runs on an interface in accordance with the ISO/IEC 14443 international standard, amongst others. These standards intend interoperability between different countries and different manufacturers of passport books.

By this, by the pressure from the US on other countries, and by the concentration on a few service providers for security technology, I would discard that there are technical incompatibilities that render it impossible for the US authorities to use the fingerprint data on the passport.

Thus I'm very skeptical that the guy in the story was not caught primarily by the facial recognition software and not by the fingerprint scan (or simply by the security officer).

QUESTIONS: Do you have any information if the fingerprint data contained on EU passports is or is not compared to the actual fingerprints when entering the US? What would be the reasons for having such an infrastructure but not making use of it? Why would reading the fingerprints on the passport be less reliable than using only the photograph on the passport?

KlausN
  • 710
  • 1
  • 5
  • 13
  • [Welcome to Skeptics!](http://meta.skeptics.stackexchange.com/questions/1505/welcome-to-new-users) Your speculation that other methods might have also found him seems irrelevant - no-one is denying that. What is the specific claim in the article that you are doubtful about? – Oddthinking Aug 25 '18 at 19:19
  • 2
    I doubt about this: "You can much more accurately identify who an individual is. You can much more rapidly speed them through processes,". I'm convinced the guy in the article was caught because of the fingerprints not because of the facial recognition. There are 23 million people coming through Dulles airport every year. The software has 99% accuracy (as claimed in the article) so that roughly 1800 people on 3 days were wrongly recognized as somebody else. This sounds like a mess, and I believe that authorities need to justify themselves by proclaiming this "success". – KlausN Aug 25 '18 at 20:54
  • I am trying to working out how to focus on that claim. I think you are making a few false assumptions that are clouding the issue. (1) "wrongly recognised as somebody else" isn't accurate. Assuming sensitivity=specificity, 1800 honest travellers would be flagged as "can't match" (and 1 in 100 impostors would get through unnoticed.) – Oddthinking Aug 26 '18 at 03:13
  • 2
    (2) French passports have biometrics that include a digital image of the face (not just fingerprints). In a 1 in 100 miss rate was acceptable, you could automate the verification to just use a face, and divert those 1800 non-matches to a fingerprint pad or for manual processing. – Oddthinking Aug 26 '18 at 03:14
  • @KlausNiederkrüger - Re *I am skeptical that facial comparison technology is an improvement in speed and accuracy over existing technologies, such as fingerprints stored on biometric passports* and your subsequent comment *I'm convinced the guy in the article was caught because of the fingerprints not because of the facial recognition.* This is your opinion; you need to make this a "notable claim." There's a big problem with these statements: US biometric passports do not contain fingerprint scans (continued). – David Hammen Aug 26 '18 at 13:06
  • Fingerprinting is a secondary defense against false entry. The US has long used facial recognition as the first line of defense against false entry. The primary facial recognition devices still used at most ports of entry are US Customs agents: i.e., people. The issue then is whether these new mechanized facial recognition devices are faster / more accurate than are those old-fashioned facial recognition devices. – David Hammen Aug 26 '18 at 13:11
  • 1
    @DavidHammen: But the guy in the story had a French passport which does contain fingerprint scans and as a non-US citizen the authorities do scan your fingerprints when entering the US. So the fingerprints should have caught this person. I mean, I could imagine that a person travels with a stolen passport of somebody who looks similar to him, but how would he manage to find a stolen passport of somebody that has almost identical fingerprints to his? – KlausN Aug 26 '18 at 15:18
  • 1
    Even if French passports do contain fingerprints, and US authorities do scan the fingerprints of non-citizens, I wouldn't take it for granted that they actually compare the one to the other. Can we find confirmation of that? I understood that the US collected fingerprints mainly for their own records, and so that passengers could be identified as the same person on a second entry. – Nate Eldredge Aug 26 '18 at 17:20
  • 1
    @NateEldredge: I was taking it for granted that the fingerprint data on the passport is compared to the fingerprint taken at the airport. For what reason would the DHS have chosen not to compare them, if they have both of them? The biometric passports seem to be using a standard file format... the only reason I can come up with is that fingerprint comparison is not sufficiently reliable and would produce too many problems, but I never heard anything like this. – KlausN Aug 26 '18 at 21:52
  • 1
    Well, I don't think it's so much about whether they would "choose" to compare them, as whether they have set up the technology to do so. It means comparing data from two different systems, and things like that are often much harder and slower to implement than people would guess - especially when government agencies are involved. There also could be legal or regulatory issues that we don't know about. My point is, let's not guess or make assumptions; let's try to find evidence. – Nate Eldredge Aug 26 '18 at 22:02
  • It is hard to find concrete information. The [wikipedia article on biometric passports][https://en.wikipedia.org/wiki/Biometric_passport] is quite technical, but it states that the technology "runs on an interface in accordance with the ISO/IEC 14443 international standard, amongst others. These standards intend interoperability between different countries and different manufacturers of passport books." – KlausN Aug 26 '18 at 22:28
  • Hello and welcome to Skeptics SE. A friendly reminder: being a **skeptic** is not the same as being a **doubter/denier**. Skeptics SE exists to perform skepticism on notable claims. It does not exist to give doubters and deniers ammunition to shoot down notable claims. If the claim is "facial recognition was used and that technique brought a suspect to the attention of security personnel", then **that** is the claim that will be examined. What will **not** be examined are questions of the sort "Yeah well why did they not do it in any other way?". That is outside the scope of Skeptics SE. –  Aug 28 '18 at 11:08
  • 1
    @KlausNiederkrüger: [ISO/IEC 14443](https://en.wikipedia.org/wiki/ISO/IEC_14443) describes the protocol for communicating with the chips. I don't think there's any doubt that US border officers are able to retrieve the data from the passport chip. My question is whether they are able to compare that data with the fingerprint scans that they collect. – Nate Eldredge Aug 28 '18 at 16:05
  • @MichaelK if the new technology identified the mismatch between the passport and the person presenting it, but old technology that has been in place for years *also* identified the mismatch, then one can argue that the new technology was not responsible for the detection. – phoog Oct 23 '18 at 16:02

1 Answers1

1

According to Facial-recognition technology at Dulles catches 3 impostors entering US:

On Monday, a woman on a flight from Accra, Ghana, presented a U.S. passport at Dulles, but CBP officials say when the facial recognition technology found a mismatch, it was determined she was actually a citizen of Cameroon. That case is now under investigation.

Two similar cases have been reported since August. A Congolese man was caught using a French passport Aug. 22; a Ghanaian woman using a U.S. passport, Sept. 8.

Sapp said the facial recognition system is used when people enter the country. “When a foreign visitor walks up to our inspection booth, the first thing we do is we take a photograph of that individual.” That photo is matched against the photo on the passport chip or the travel manifest. “We’ll know right there, in matter of seconds, whether that person in front of us, with the travel document, is the authentic holder of that travel document.”

Community
  • 1
DavePhD
  • 103,432
  • 24
  • 436
  • 464
  • "When a foreign visitor walks up to our inspection booth, the first thing we do is we take a photograph of that individual": apparently they also photograph US citizens, too. Otherwise they would be unable to use facial recognition software to confirm that a US passport matches the person presenting it. But they certainly don't fingerprint US citizens, so this does confirm the claim in the question. – phoog Oct 23 '18 at 15:59