Tag Archives: facial recognition

If We Don’t Train Facial Recognition Users, There Will Be No Facial Recognition

(Part of the biometric product marketing expert series)

We get all sorts of great tools, but do we know how to use them? And what are the consequences if we don’t know how to use them? Could we lose the use of those tools entirely due to bad publicity from misuse?

Hida Viloria. By Intersex77 – Own work, CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=98625035

Do your federal facial recognition users know what they are doing?

I recently saw a WIRED article that primarily talked about submitting Parabon Nanolabs-generated images to a facial recognition program. But buried in the article was this alarming quote:

According to a report released in September by the US Government Accountability Office, only 5 percent of the 196 FBI agents who have access to facial recognition technology from outside vendors have completed any training on how to properly use the tools.

From https://www.wired.com/story/parabon-nanolabs-dna-face-models-police-facial-recognition/

Now I had some questions after reading that sentence: namely, what does “have access” mean? To answer those questions, I had to find the study itself, GAO-23-105607, Facial Recognition Services: Federal Law Enforcement Agencies Should Take Actions to Implement Training, and Policies for Civil Liberties.

It turns out that the study is NOT limited to FBI use of facial recognition services, but also addresses six other federal agencies: the Bureau of Alcohol, Tobacco, Firearms and Explosives (the guvmint doesn’t believe in the Oxford comma); U.S. Customs and Border Protection; the Drug Enforcement Administration; Homeland Security Investigations; the U.S. Marshals Service; and the U.S. Secret Service.

In addition, the study confines itself to four facial recognition services: Clearview AI, IntelCenter, Marinus Analytics, and Thorn. It does not address other uses of facial recognition by the agencies, such as the FBI’s use of IDEMIA in its Next Generation Identification system (IDEMIA facial recognition technology is also used by the Department of Defense).

Two of the GAO’s findings:

  • Initially, none of the seven agencies required users to complete facial recognition training. As of April 2023, two of the agencies (Homeland Security Investigations and the U.S. Marshals Service) required training, two (the FBI and Customs and Border Protection) did not, and the other three had quit using these four facial recognition services.
  • The FBI stated that facial recognition training was recommended as a “best practice,” but not mandatory. And when something isn’t mandatory, you can guess what happened:

GAO found that few of these staff completed the training, and across the FBI, only 10 staff completed facial recognition training of 196 staff that accessed the service. FBI said they intend to implement a training requirement for all staff, but have not yet done so. 

From https://www.gao.gov/products/gao-23-105607.

So if you use my three levels of importance (TLOI) model, facial recognition training is important, but not critically important. Therefore, it wasn’t done.

The detailed version of the report includes additional information on the FBI’s training requirements…I mean recommendations:

Although not a requirement, FBI officials said they recommend (as
a best practice) that some staff complete FBI’s Face Comparison and
Identification Training when using Clearview AI. The recommended
training course, which is 24 hours in length, provides staff with information on how to interpret the output of facial recognition services, how to analyze different facial features (such as ears, eyes, and mouths), and how changes to facial features (such as aging) could affect results.

From https://www.gao.gov/assets/gao-23-105607.pdf.

However, this type of training was not recommended for all FBI users of Clearview AI, and was not recommended for any FBI users of Marinus Analytics or Thorn.

I should note that the report was issued in September 2023, based upon data gathered earlier in the year, and that for all I know the FBI now mandates such training.

Or maybe it doesn’t.

What about your state and local facial recognition users?

Of course, training for federal facial recognition users is only a small part of the story, since most of the law enforcement activity takes place at the state and local level. State and local users need training so that they can understand:

  • The anatomy of the face, and how it affects comparisons between two facial images.
  • How cameras work, and how this affects comparisons between two facial images.
  • How poor quality images can adversely affect facial recognition.
  • How facial recognition should ONLY be used as an investigative lead.

If state and local users received this training, none of the false arrests over the last few years would have taken place.

What are the consequences of no training?

Could I repeat that again?

If facial recognition users had been trained, none of the false arrests over the last few years would have taken place.

  • The users would have realized that the poor images were not of sufficient quality to determine a match.
  • The users would have realized that even if they had been of sufficient quality, facial recognition must only be used as an investigative lead, and once other data had been checked, the cases would have fallen apart.

But the false arrests gave the privacy advocates the ammunition they needed.

Not to insist upon proper training in the use of facial recognition.

But to ban the use of facial recognition entirely.

Like nuclear or biological weapons, facial recognition’s threat to human society and civil liberties far outweighs any potential benefits. Silicon Valley lobbyists are disingenuously calling for regulation of facial recognition so they can continue to profit by rapidly spreading this surveillance dragnet. They’re trying to avoid the real debate: whether technology this dangerous should even exist. Industry-friendly and government-friendly oversight will not fix the dangers inherent in law enforcement’s discriminatory use of facial recognition: we need an all-out ban.

From https://www.banfacialrecognition.com/

(And just wait until the anti-facial recognition forces discover that this is not only a plot of evil Silicon Valley, but also a plot of evil non-American foreign interests located in places like Paris and Tokyo.)

Because the anti-facial recognition forces want us to remove the use of technology and go back to the good old days…of eyewitness misidentification.

Eyewitness misidentification contributes to an overwhelming majority of wrongful convictions that have been overturned by post-conviction DNA testing.

Eyewitnesses are often expected to identify perpetrators of crimes based on memory, which is incredibly malleable. Under intense pressure, through suggestive police practices, or over time, an eyewitness is more likely to find it difficult to correctly recall details about what they saw. 

From https://innocenceproject.org/eyewitness-misidentification/.

And these people don’t stay in jail for a night or two. Some of them remain in prison for years until the eyewitness misidentification is reversed.

Archie Williams moments after his exoneration on March 21, 2019. Photo by Innocence Project New Orleans. From https://innocenceproject.org/fingerprint-database-match-establishes-archie-williams-innocence/

Eyewitnesses, unlike facial recognition algorithms, cannot be tested for accuracy or bias.

And if we don’t train facial recognition users in the technology, then we’re going to lose it.

Identification Perfection is Impossible

(Part of the biometric product marketing expert series)

There are many different types of perfection.

Jehan Cauvin (we don’t spell his name like he spelled it). By Titian – Bridgeman Art Library: Object 80411, Public Domain, https://commons.wikimedia.org/w/index.php?curid=6016067

This post concentrates on IDENTIFICATION perfection, or the ability to enjoy zero errors when identifying individuals.

The risk of claiming identification perfection (or any perfection) is that a SINGLE counter-example disproves the claim.

  • If you assert that your biometric solution offers 100% accuracy, a SINGLE false positive or false negative shatters the assertion.
  • If you claim that your presentation attack detection solution exposes deepfakes (face, voice, or other), then a SINGLE deepfake that gets past your solution disproves your claim.
  • And as for the pre-2009 claim that latent fingerprint examiners never make a mistake in an identification…well, ask Brandon Mayfield about that one.

In fact, I go so far as to avoid using the phrase “no two fingerprints are alike.” Many years ago (before 2009) in an International Association for Identification meeting, I heard someone justify the claim by saying, “We haven’t found a counter-example yet.” That doesn’t mean that we’ll NEVER find one.

You’ve probably heard me tell the story before about how I misspelled the word “quality.”

In a process improvement document.

While employed by Motorola (pre-split).

At first glance, it appears that Motorola would be the last place to make a boneheaded mistake like that. After all, Motorola is known for its focus on quality.

But in actuality, Motorola was the perfect place to make such a mistake, since it was one of the champions of the “Six Sigma” philosophy (which targets a maximum of 3.4 defects per million opportunities). Motorola realized that manufacturing perfection is impossible, so manufacturers (and the people in Motorola’s weird Biometric Business Unit) should instead concentrate on reducing the error rate as much as possible.

So one misspelling could be tolerated, but I shudder to think what would have happened if I had misspelled “quality” a second time.

Login.gov and IAL2 #realsoonnow

Back in August 2023, the U.S. General Services Administration published a blog post that included the following statement:

Login.gov is on a path to providing an IAL2-compliant identity verification service to its customers in a responsible, equitable way. Building on the strong evidence-based identity verification that Login.gov already offers, Login.gov is on a path to providing IAL2-compliant identity verification that ensures both strong security and broad and equitable access.

From https://www.gsa.gov/blog/2023/08/18/reducing-fraud-and-increasing-access-drives-record-adoption-and-usage-of-logingov

It’s nice to know…NOW…that Login.gov is working to achieve IAL2.

This post explains what the August 2023 GSA post said, and what it didn’t say.

But first, I’ll define what Login.gov and “IAL2” are.

What is Login.gov?

From https://login.gov/

Here is what Login.gov says about itself:

Login.gov is a secure sign in service used by the public to sign in to participating government agencies. Participating agencies will ask you to create a Login.gov account to securely access your information on their website or application.

You can use the same username and password to access any agency that partners with Login.gov. This streamlines your process and eliminates the need to remember multiple usernames and passwords.

From https://www.login.gov/what-is-login/
From https://www.gsa.gov/about-us/organization/federal-acquisition-service/technology-transformation-services

Obviously there are a number of private companies (over 80 last I counted) that provide secure access to information, but Login.gov is provided by the government itself—specifically by the General Services Administration’s Technology Transformation Services. Agencies at the federal, state, and local level can work with the GSA TTS’ “18F” organization to implement solutions such as Login.gov.

Why would agencies implement Login.gov? Because the agencies want to protect their constituents’ information. If fraudsters capture personally identifiable information (PII) of someone applying for government services, the breached government agency will face severe repurcussions. Login.gov is supposed to protect its partner agencies from these nightmares.

How does Login.gov do this?

  • Sometimes you might use two-factor authentication consisting of a password and a second factor such as an SMS code or the use of an authentication app.
  • In more critical cases, Login.gov requests a more reliable method of identification, such as a government-issued photo ID (driver’s license, passport, etc.).

What is IAL2?

At the risk of repeating myself, I’ll briefly go over what “Identity Assurance Level 2” (IAL2) is.

From https://pages.nist.gov/800-63-3/sp800-63a.html

The U.S. National Institute of Standards and Technology, in its publication NIST SP 800-63a, has defined “identity assurance levels” (IALs) that can be used when dealing with digital identities. It’s helpful to review how NIST has defined the IALs. (I’ll define the other acronyms as we go along.)

Assurance in a subscriber’s identity is described using one of three IALs:

IAL1: There is no requirement to link the applicant to a specific real-life identity. Any attributes provided in conjunction with the subject’s activities are self-asserted or should be treated as self-asserted (including attributes a [Credential Service Provider] CSP asserts to an [Relying Party] RP). Self-asserted attributes are neither validated nor verified.

IAL2: Evidence supports the real-world existence of the claimed identity and verifies that the applicant is appropriately associated with this real-world identity. IAL2 introduces the need for either remote or physically-present identity proofing. Attributes could be asserted by CSPs to RPs in support of pseudonymous identity with verified attributes. A CSP that supports IAL2 can support IAL1 transactions if the user consents.

IAL3: Physical presence is required for identity proofing. Identifying attributes must be verified by an authorized and trained CSP representative. As with IAL2, attributes could be asserted by CSPs to RPs in support of pseudonymous identity with verified attributes. A CSP that supports IAL3 can support IAL1 and IAL2 identity attributes if the user consents.

From https://pages.nist.gov/800-63-3/sp800-63a.html#sec2

So in its simplest terms, IAL2 requires evidence of a verified credential so that an online person can be linked to a real-life identity. If someone says they’re “John Bredehoft” and fills in an online application to receive government services, IAL2 compliance helps to ensure that the person filling out the online application truly IS John Bredehoft, and not Bernie Madoff.

As more and more of us conduct business—including government business—online, IAL2 compliance is essential to reduce fraud.

One more thing about IAL2 compliance. The mere possession of a valid government issued photo ID is NOT sufficient for IAL2 compliance. After all, Bernie Madoff may be using John Bredehoft’s driver’s license. To make sure that it’s John Bredehoft using John Bredehoft’s driver’s license, an additional check is needed.

This has been explained by ID.me, a private company that happens to compete with Login.gov to provide identity proofing services to government agencies.

Biometric comparison (e.g., selfie with liveness detection or fingerprint) of the strongest piece of evidence to the applicant

From https://network.id.me/article/what-is-nist-ial2-identity-verification/

So you basically take the information on a driver’s license and perform a facial recognition 1:1 comparison with the person possessing the driver’s license, ideally using liveness detection, to make sure that the presented person is not a fake.

From https://www.nist.gov/news-events/news/2023/09/whats-wrong-picture-nist-face-analysis-program-helps-find-answers

So what?

So the GSA was apparently claiming how secure Login.gov was. Guess who challenged the claim?

The GSA.

Now sometimes it’s ludicrous to think that the government can police itself, but in some cases government actually identifies government faults.

Of course, this works best when you can identify problems with some other government entity.

Which is why the General Services Administration has an Inspector General. And in March 2023, the GSA Inspector General released a report with the following title: “GSA Misled Customers on Login.gov’s Compliance with Digital Identity Standards.”

From the GSA Inspector General.

The title is pretty clear, but Fedscoop summarized the findings for those who missed the obvious:

As part of an investigation that has run since last April (2022), GSA’s Office of the Inspector General found that the agency was billing agencies for IAL2-compliant services, even though Login.gov did not meet Identity Assurance Level 2 (IAL2) standards.

GSA knowingly billed over $10 million for services provided through contracts with other federal agencies, even though Login.gov is not IAL2 compliant, according to the watchdog.

From https://fedscoop.com/gsa-login-gov-watchdog-report/

So now GSA is explicitly saying that Login.gov ISN’T IAL2-compliant.

Which helps its private sector competitors.

Clean Data is the New Oxygen, and Dirty Data is the New Carbon Monoxide

I have three questions for you, but don’t sweat; I’m giving you the answers.

  1. How long can you survive without pizza? Years (although your existence will be hellish).
  2. OK, how long can you survive without water? From 3 days to 7 days.
  3. OK, how long can you survive without oxygen? Only 10 minutes.

This post asks how long a 21st century firm can survive without data, and what can happen if the data is “dirty.”

How does Mika survive?

Have you heard of Mika? Here’s her LinkedIn profile.

From Mika’s LinkedIn profile at https://www.linkedin.com/in/mika-ai-ceo/

Yes, you already know that I don’t like LinkedIn profiles that don’t belong to real people. But this one is a bit different.

Mika is the Chief Executive Officer of Dictador, a Polish-Colombian spirits firm, and is responsible for “data insight, strategic provocation and DAO community liaison.” Regarding data insight, Mika described her approach in an interview with Inside Edition:

My decision making process relies on extensive data analysis and aligning with the company’s strategic objectives. It’s devoid of personal bias ensuring unbiased and strategic choices that prioritize the organization’s best interests.

From the transcript to https://www.youtube.com/watch?v=8BQEyQ2-awc
From https://www.youtube.com/watch?v=8BQEyQ2-awc

Mika was brought to my attention by accomplished product marketer/artist Danuta (Dana) Deborgoska. (She’s appeared in the Bredemarket blog before, though not by name.) Dana is also Polish (but not Colombian) and clearly takes pride in the artificial intelligence accomplishments of this Polish-headquartered company. You can read her LinkedIn post to see her thoughts, one of which was as follows:

Data is the new oxygen, and we all know that we need clean data to innovate and sustain business models.

From Dana Debogorska’s LinkedIn post.

Dana succinctly made two points:

  1. Data is the new oxygen.
  2. We need clean data.

Point one: data is the new oxygen

There’s a reference to oxygen again, but it’s certainly appropriate. Just as people cannot survive without oxygen, Generative AI cannot survive without data.

But the need for data predates AI models. From 2017:

Reliance Industries Chairman Mukesh Ambani said India is poised to grow…but to make that happen the country’s telecoms and IT industry would need to play a foundational role and create the necessary digital infrastructure.

Calling data the “oxygen” of the digital economy, Ambani said the telecom industry had the urgent task of empowering 1.3 billion Indians with the tools needed to flourish in the digital marketplace.

From India Times.

And we can go back centuries in history and find examples when a lack of data led to catastrophe. Like the time in 1776 when the Hessians didn’t know that George Washington and his troops had crossed the Delaware.

Public domain. From https://www.metmuseum.org/art/collection/search/11417

Point two: we need clean data

Of course, the presence or absence of data alone is not enough. As Debogorska notes, we don’t just need any data; we need CLEAN data, without error and without bias. Dirty data is like carbon monoxide, and as you know carbon monoxide is harmful…well, most of the time.

That’s been the challenge not only with artificial intelligence, but with ALL aspects of data gathering.

The all-male board of directors of a fertilizer company in 1960. Fair use. From the New York Times.

In all of these cases, someone (Amazon, Enron’s shareholders, or NIST) asked questions about the cleanliness of the data, and then set out to answer those questions.

  • In the case of Amazon’s recruitment tool and the company Enron, the answers caused Amazon to abandon the tool and Enron to abandon its existence.
  • Despite the entreaties of so-called privacy advocates (who prefer the privacy nightmare of physical driver’s licenses to the privacy-preserving features of mobile driver’s licenses), we have not abandoned facial recognition, but we’re definitely monitoring it in a statistical (not an anecdotal) sense.

The cleanliness of the data will continue to be the challenge as we apply artificial intelligence to new applications.

Clean room of a semiconductor manufacturing facility. Uploaded by Duk 08:45, 16 Feb 2005 (UTC) – http://www.grc.nasa.gov/WWW/ictd/content/labmicrofab.html (original) and https://images.nasa.gov/details/GRC-1998-C-01261 (high resolution), Public Domain, https://commons.wikimedia.org/w/index.php?curid=60825

Point three: if you’re not saying things, then you’re not selling

(Yes, this is the surprise point.)

Dictador is talking about Mika.

Are you talking about your product, or are you keeping mum about it?

I have more to…um…say about this. Follow this link.

Pangiam May Be Acquired Next Year

Things change. Pangiam, a company that didn’t even exist a few years ago, and that started off by acquiring a one-off project from a local government agency, is now itself a friendly acquisition target (pending stockholder and regulatory approvals).

From MWAA to Pangiam

Back when I worked for IDEMIA and helped to market its border control solutions, one of our competitors for airport business was an airport itself—specifically, the Metropolitan Washington Airports Authority. Rather than buying a biometric exit solution from someone else, the MWAA developed its own, called veriScan.

2021 image from the former airportveriscan website.

After I left IDEMIA, the MWAA decided that it didn’t want to be in the software business any more, and sold veriScan to a new company, Pangiam. I posted about this decision and the new company in this blog.

ALEXANDRIA, Va., March 19, 2021 /PRNewswire/ — Pangiam, a technology-based security and travel services provider, announced today that it has acquired veriScan, an integrated biometric facial recognition system for airports and airlines, from the Metropolitan Washington Airports Authority (“Airports Authority”). Terms of the transaction were not disclosed.

From PR Newswire.

But Pangiam was just getting started.

Trueface, FRTE, stadiums, and artificial intelligence

Results for the NIST FRTE 1:N pangiam-000 algorithm, captured November 6, 2023 from NIST.

A few months later Pangiam acquired Trueface and therefore earned a spot on the NIST FRTE 1:N (formerly FRVT 1:N) rankings and an interest in the stadium/venue identity verification/authentication market.

By Chris6d – Own work, CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=101751795

Meanwhile Pangiam continued to build up its airport business and also improved its core facial recognition technology.

After that I personally concentrated on other markets, and therefore missed the announcements of Pangiam Bridge (introducing artificial intelligence into Pangiam’s border crossing offering) and Project DARTMOUTH (devoted to using artificial intelligence and pattern analysis to airline baggage, cargo, and shipments).

So what will Pangiam work on next? Where will it expand? What will it acquire?

Nothing.

Enter BigBear.ai

Pangiam itself is now an acquisition target.

COLUMBIA, MD.— November 6, 2023 — BigBear.ai (NYSE: BBAI), a leading provider of AI-enabled business intelligence solutions, today announced a definitive merger agreement to acquire Pangiam Intermediate Holdings, LLC (Pangiam), a leader in Vision AI for the global trade, travel, and digital identity industries, for approximately $70 million in an all-stock transaction. The combined company will create one of the industry’s most comprehensive Vision AI portfolios, combining Pangiam’s facial recognition and advanced biometrics with BigBear.ai’s computer vision capabilities, positioning the company as a foundational leader in one of the fastest growing categories for the application of AI. The proposed acquisition is expected to close in the first quarter of 2024, subject to customary closing conditions, including approval by the holders of a majority of BigBear.ai’s outstanding common shares and receipt of regulatory approval.

From bigbear.ai.

Yet another example of how biometrics is now just a minor part of general artificial intelligence efforts. Identify a face or a grenade, it’s all the same.

Anyway, let’s check back in a few months. Because of the technology involved, this proposed acquisition will DEFINITELY merit government review.

Converting Prospects For Your Firm’s “Something You Are” Solution

As identity/biometric professionals well know, there are five authentication factors that you can use to gain access to a person’s account. (You can also use these factors for identity verification to establish the person’s account in the first place.)

I described one of these factors, “something you are,” in a 2021 post on the five authentication factors.

Something You Are. I’ve spent…a long time with this factor, since this is the factor that includes biometrics modalities (finger, face, iris, DNA, voice, vein, etc.). It also includes behavioral biometrics, provided that they are truly behavioral and relatively static.

From https://bredemarket.com/2021/03/02/the-five-authentication-factors/

As I mentioned in August, there are a number of biometric modalities, including face, fingerprint, iris, hand geometry, palm print, signature, voice, gait, and many more.

From Sandeep Kumar, A. Sony, Rahul Hooda, Yashpal Singh, in Journal of Advances and Scholarly Researches in Allied Education | Multidisciplinary Academic Research, “Multimodal Biometric Authentication System for Automatic Certificate Generation.”

If your firm offers an identity solution that partially depends upon “something you are,” then you need to create content (blog, case study, social media, white paper, etc.) that converts prospects for your identity/biometric product/service and drives content results.

Bredemarket can help.

Click below for details.

Drive Content Results with Bredemarket Identity Firm Services.

The Imperfect Way to Enforce New York’s Child Data Protection Act

It’s often good to use emotion in your marketing.

For example, when biometric companies want to justify the use of their technology, they have found that it is very effective to position biometrics as a way to combat sex trafficking.

Similarly, moves to rein in social media are positioned as a way to preserve mental health.

By Marc NL at English Wikipedia – Transferred from en.wikipedia to Commons., Public Domain, https://commons.wikimedia.org/w/index.php?curid=2747237

Now that’s a not-so-pretty picture, but it effectively speaks to emotions.

“If poor vulnerable children are exposed to addictive, uncontrolled social media, YOUR child may end up in a straitjacket!”

In New York state, four government officials have declared that the ONLY way to preserve the mental health of underage social media users is via two bills, one of which is the “New York Child Data Protection Act.”

But there is a challenge to enforce ALL of the bill’s provisions…and only one way to solve it. An imperfect way—age estimation.

This post only briefly addresses the alleged mental health issues of social media before plunging into one of the two proposed bills to solve the problem. It then examines a potentially unenforceable part of the bill and a possible solution.

Does social media make children sick?

Letitia “Tish” James is the 67th Attorney General for the state of New York. From https://ag.ny.gov/about/meet-letitia-james

On October 11, a host of New York State government officials, led by New York State Attorney General Letitia James, jointly issued a release with the title “Attorney General James, Governor Hochul, Senator Gounardes, and Assemblymember Rozic Take Action to Protect Children Online.”

Because they want to protect the poor vulnerable children.

By Paolo Monti – Available in the BEIC digital library and uploaded in partnership with BEIC Foundation.The image comes from the Fondo Paolo Monti, owned by BEIC and located in the Civico Archivio Fotografico of Milan., CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=48057924

And because the major U.S. social media companies are headquartered in California. But I digress.

So why do they say that children need protection?

Recent research has shown devastating mental health effects associated with children and young adults’ social media use, including increased rates of depression, anxiety, suicidal ideation, and self-harm. The advent of dangerous, viral ‘challenges’ being promoted through social media has further endangered children and young adults.

From https://ag.ny.gov/child-online-safety

Of course one can also argue that social media is harmful to adults, but the New Yorkers aren’t going to go that far.

So they are just going to protect the poor vulnerable children.

CC BY-SA 4.0.

This post isn’t going to deeply analyze one of the two bills the quartet have championed, but I will briefly mention that bill now.

  • The “Stop Addictive Feeds Exploitation (SAFE) for Kids Act” (S7694/A8148) defines “addictive feeds” as those that are arranged by a social media platform’s algorithm to maximize the platform’s use.
  • Those of us who are flat-out elderly vaguely recall that this replaced the former “chronological feed” in which the most recent content appeared first, and you had to scroll down to see that really cool post from two days ago. New York wants the chronological feed to be the default for social media users under 18.
  • The bill also proposes to limit under 18 access to social media without parental consent, especially between midnight and 6:00 am.
  • And those who love Illinois BIPA will be pleased to know that the bill allows parents (and their lawyers) to sue for damages.

Previous efforts to control underage use of social media have faced legal scrutinity, but since Attorney General James has sworn to uphold the U.S. Constitution, presumably she has thought about all this.

Enough about SAFE for Kids. Let’s look at the other bill.

The New York Child Data Protection Act

The second bill, and the one that concerns me, is the “New York Child Data Protection Act” (S7695/A8149). Here is how the quartet describes how this bill will protect the poor vulnerable children.

CC BY-SA 4.0.

With few privacy protections in place for minors online, children are vulnerable to having their location and other personal data tracked and shared with third parties. To protect children’s privacy, the New York Child Data Protection Act will prohibit all online sites from collecting, using, sharing, or selling personal data of anyone under the age of 18 for the purposes of advertising, unless they receive informed consent or unless doing so is strictly necessary for the purpose of the website. For users under 13, this informed consent must come from a parent.

From https://ag.ny.gov/child-online-safety

And again, this bill provides a BIPA-like mechanism for parents or guardians (and their lawyers) to sue for damages.

But let’s dig into the details. With apologies to the New York State Assembly, I’m going to dig into the Senate version of the bill (S7695). Bear in mind that this bill could be amended after I post this, and some of the portions that I cite could change.

From https://www.nysenate.gov/legislation/bills/2023/S7695

The “definitions” section of the bill includes the following:

“MINOR” SHALL MEAN A NATURAL PERSON UNDER THE AGE OF EIGHTEEN.

From https://www.nysenate.gov/legislation/bills/2023/S7695, § 899-EE, 2.

This only applies to natural persons. So the bots are safe, regardless of age.

Speaking of age, the age of 18 isn’t the only age referenced in the bill. Here’s a part of the “privacy protection by default” section:

§ 899-FF. PRIVACY PROTECTION BY DEFAULT.

1. EXCEPT AS PROVIDED FOR IN SUBDIVISION SIX OF THIS SECTION AND SECTION EIGHT HUNDRED NINETY-NINE-JJ OF THIS ARTICLE, AN OPERATOR SHALL NOT PROCESS, OR ALLOW A THIRD PARTY TO PROCESS, THE PERSONAL DATA OF A COVERED USER COLLECTED THROUGH THE USE OF A WEBSITE, ONLINE SERVICE, ONLINE APPLICATION, MOBILE APPLICA- TION, OR CONNECTED DEVICE UNLESS AND TO THE EXTENT:

(A) THE COVERED USER IS TWELVE YEARS OF AGE OR YOUNGER AND PROCESSING IS PERMITTED UNDER 15 U.S.C. § 6502 AND ITS IMPLEMENTING REGULATIONS; OR

(B) THE COVERED USER IS THIRTEEN YEARS OF AGE OR OLDER AND PROCESSING IS STRICTLY NECESSARY FOR AN ACTIVITY SET FORTH IN SUBDIVISION TWO OF THIS SECTION, OR INFORMED CONSENT HAS BEEN OBTAINED AS SET FORTH IN SUBDIVISION THREE OF THIS SECTION.

From https://www.nysenate.gov/legislation/bills/2023/S7695

So a lot of this bill depends upon whether a person is over or under the age of eighteen, or over or under the age of thirteen.

And that’s a problem.

How old are you?

The bill needs to know whether or not a person is 18 years old. And I don’t think the quartet will be satisfied with the way that alcohol websites determine whether someone is 21 years old.

From https://www.budlight.com/

This age verification method is…not that robust.

Attorney General James and the others would presumably prefer that the social media companies verify ages with a government-issued ID such as a state driver’s license, a state identification card, or a national passport. This is how most entities verify ages when they have to satisfy legal requirements.

For some people, even some minors, this is not that much of a problem. Anyone who wants to drive in New York State must have a driver’s license, and you have to be at least 16 years old to get a driver’s license. Admittedly some people in the city never bother to get a driver’s license, but at some point these people will probably get a state ID card.

You don’t need a driver’s license to ride the New York City subway, but if the guitarist wants to open a bank account for his cash it would help him prove his financial identity. By David Shankbone – Own work, CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=2639495
  • However, there are going to be some 17 year olds who don’t have a driver’s license, government ID or passport.
  • And some 16 year olds.
  • And once you look at younger people—15 year olds, 14 year olds, 13 year olds, 12 year olds—the chances of them having a government-issued identification document are much less.

What are these people supposed to do? Provide a birth certificate? And how will the social media companies know if the birth certificate is legitimate?

Birth certificate of a B.H. Obama II. From https://obamawhitehouse.archives.gov/sites/default/files/rss_viewer/birth-certificate-long-form.pdf

But there’s another way to determine ages—age estimation.

How old are you, part 2

As long-time readers of the Bredemarket blog know, I have struggled with the issue of age verification, especially for people who do not have driver’s licenses or other government identification. Age estimation in the absence of a government ID is still an inexact science, as even Yoti has stated.

Our technology is accurate for 6 to 12 year olds, with a mean absolute error (MAE) of 1.3 years, and of 1.4 years for 13 to 17 year olds. These are the two age ranges regulators focus upon to ensure that under 13s and 18s do not have access to age restricted goods and services.

From https://www.yoti.com/wp-content/uploads/Yoti-Age-Estimation-White-Paper-March-2023.pdf

So if a minor does not have a government ID, and the social media firm has to use age estimation to determine a minor’s age for purposes of the New York Child Data Protection Act, the following two scenarios are possible:

  • An 11 year old may be incorrectly allowed to give informed consent for purposes of the Act.
  • A 14 year old may be incorrectly denied the ability to give informed consent for purposes of the Act.

Is age estimation “good enough for government work”?

Why Age-Restricted Gig Economy Companies Need Continuous Authentication (and Liveness Detection)

If you ask any one of us in the identity verification industry, we’ll tell you how identity verification proves that you know who is accessing your service.

  • During the identity verification/onboarding step, one common technique is to capture the live face of the person who is being onboarded, then compare that to the face captured from the person’s government identity document. As long as you have assurance that (a) the face is live and not a photo, and (b) the identity document has not been tampered, you positively know who you are onboarding.
  • The authentication step usually captures a live face and compares it to the face that was captured during onboarding, thus positively showing that the right person is accessing the previously onboarded account.

Sound like the perfect solution, especially in industries that rely on age verification to ensure that people are old enough to access the service.

Therefore, if you are employing robust identity verification and authentication that includes age verification, this should never happen.

By LukaszKatlewa – Own work, CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=49248622

Eduardo Montanari, who manages delivery logistics at a burger shop north of São Paulo, has noticed a pattern: Every time an order pickup is assigned to a female driver, there’s a good chance the worker is a minor.

From https://restofworld.org/2023/underage-gig-workers-brazil/

An underage delivery person who has been onboarded and authenticated, and whose age has been verified? That’s impossible, you say! Read on.

31,000 people already know how to bypass onboarding and authentication

Rest of World wrote an article (tip of the hat to Bianca Gonzalez of Biometric Update) entitled “Underage gig workers keep outsmarting facial recognition.”

Outsmarting onboarding

How do the minors do it?

On YouTube, a tutorial — one of many — explains “how to deliver as a minor.” It has over 31,000 views. “You have to create an account in the name of a person who’s the right age. I created mine in my mom’s name,” says a boy, who identifies himself as a minor in the video.

From https://restofworld.org/2023/underage-gig-workers-brazil/
From https://www.youtube.com/watch?v=59vaKab4g2M. “Botei no da minha mãe não conta da minha.” (“I put it on my mother’s account, it doesn’t count on mine.”)

Once a cooperative parent or older sibling agrees to help, the account is created in the older person’s name, the older person’s face and identity document is used to create the account, and everything is valid.

Outsmarting authentication

Yes, but what about authentication?

That’s why it’s helpful to use a family member, or someone who lives in the minor’s home.

Let’s say little Maria is at home, during her homework, when her gig economy app rings with a delivery request. Now Maria was smart enough to have her older sister Irene or her mama Cecile perform the onboarding with the delivery app. If she’s at home, she can go to Irene or Cecile, have them perform the authentication, and then she’s off on her bike to make money.

(Alternatively, if the app does not support liveness detection, Maria can just hold a picture of Irene or Cecile up to the camera and authenticate.)

  • The onboarding process was completed by the account holder.
  • The authentication was completed by the account holder.
  • But the account holder isn’t the one that’s actually using the service. Once authentication is complete, anyone can access the service.

So how do you stop underage gig economy use?

According to Rest of World, one possible solution is to tattle on underage delivery people. If you see something, say something.

But what’s the incentive for a restaurant owner or delivery recipient to report that their deliveries are being performed by a kid?

By Marcello Casal Jr./Agência Brasil – http://www.agenciabrasil.gov.br/media/imagens/2008/02/20/1145mc0222a.jpg/view/normal, CC BY 2.5 br, https://commons.wikimedia.org/w/index.php?curid=3590294

“The feeling we have is that, at least this poor boy is working. I know this is horrible, but here in Brazil we end up seeing it as an opportunity … It’s ridiculous,” (psychologist Regiane Couto) said.

From https://restofworld.org/2023/underage-gig-workers-brazil/

A much better solution is to replace one-time authetication with continuous authentication, or at least be smarter in authentication. For example, a gig delivery worker could be required to authenticate at multiple points in the process:

  • When the worker receives the delivery request.
  • When the worker arrives at the restaurant.
  • When the worker makes the delivery.

It’s too difficult to drag big sister Irene or mama Cecile to ALL of these points.

This guy’s over 18. I think. By shopblocks – https://www.flickr.com/photos/155237687@N06/34720130102/, CC BY 2.0, https://commons.wikimedia.org/w/index.php?curid=89964934

As an added bonus, these authetications provide timestamps of critical points in the delivery process, which the delivery company and/or restaurant can use for their analytics.

Problem solved.

Except that little Maria doesn’t have any excuse and has to complete her homework.

Safety vs. Privacy in Montana School Video Surveillance

At the highest level, debates regarding government and enterprise use of biometric technology boil down to a debate about whether to keep people safe, or whether to preserve individual privacy.

In the state of Montana, school safety is winning over school privacy—for now.

The one exception in Montana Senate Bill 397

Biometric Update links to a Helena Independent Record article on how Montana’s far-reaching biometric ban has one significant exception.

The state Legislature earlier this year passed a law barring state and local governments from continuous use of facial recognition technology, typically in the form of cameras capable of reading and collecting a person’s biometric data, like the identifiable features of their face and body. A bipartisan group of legislators went toe-to-toe with software companies and law enforcement in getting Senate Bill 397 over the finish line, contending public safety concerns raised by the technology’s supporters don’t overcome individual privacy rights. 

School districts, however, were specifically carved out of the definition of state and local governments to which the facial recognition technology law applies.

From the Helena Independent Record.

At a minimum Montana school districts seek to abide by two existing Federal laws when installating facial recognition and video surveillance systems.

Without many state-level privacy protection laws in place, school policies typically lean on the Children’s Online Privacy Protection Act (COPPA), a federal law requiring parental consent in order for websites to collect data on their children, or the Family Educational Rights and Privacy Act (FERPA), which protects the privacy of student education records. 

From the Helena Independent Record.

If a vendor doesn’t agree to abide by these laws, then the Montana School Board Association recommends that the school district not do business with the vendor.

Other vendors agree. Here is the statement of one vendor, Verkada (you’ll see them again later) on FERPA:

The Family Educational Rights and Privacy Act was passed by the US federal government to protect the privacy of students’ educational records. This law requires public schools and school districts to give families control over any personally identifiable information about the student.

Verkada provides educational organizations the tools they need to maintain FERPA compliance, such as face blurring for archived footage.

From https://www.verkada.com/security/#compliance

Simms High School’s use of the technology

How are the schools using these systems? In ways you may expect.

From https://www.srvs.k12.mt.us/apps/pages/index.jsp?uREC_ID=356183&type=d

(The Sun River Valley School District’s) use of the technology is more focused on keeping people who shouldn’t be on school property away, he said, such as a parent who lost custody of their child.

(Simms) High School Principal Luke McKinley said it’s been more frequent to use the facial recognition technology during extra-curricular activities, when football fans get too rowdy for a high school sports event. 

From the Helena Independent Record.

Technology (in this case from Verkada) helps the Sun River School District, especially in its rural setting. Back in 2022, it took law enforcement an estimated 45 minutes to respond to school incidents. The hope is that the technology could identify those who engaged in illegal activity, or at least deter it.

What about other school districts?

When I created my educational identity page, I included the four key words “When permitted by law.” While Montana school districts are currently permitted to use facial recognition and video surveillance, other school districts need to check their local laws before implementing such a system, and also need to ensure that they comply with federal laws such as COPPA and FERPA.

I may be, um, biased in my view, but as long as the school district (or law enforcement agency, or apartment building owner, or whoever) complies with all applicable laws, and implements the technology with a primary purpose of protecting people rather than spying on them, facial recognition is a far superior tool to protect people than manual recognition methods that rely on all-too-fallible human beings.

Vision Transformer (ViT) Models and Presentation Attack Detection

I tend to view presentation attack detection (PAD) through the lens of iBeta or occasionally of BixeLab. But I need to remind myself that these are not the only entities examining PAD.

A recent paper authored by Koushik Srivatsan, Muzammal Naseer, and Karthik Nandakumar of the Mohamed Bin Zayed University of Artificial Intelligence (MBZUAI) addresses PAD from a research perspective. I honestly don’t understand the research, but perhaps you do.

Flip spoofing his natural appearance by portraying Geraldine. Some were unable to detect the attack. By NBC Television. – eBay itemphoto frontphoto back, Public Domain, https://commons.wikimedia.org/w/index.php?curid=16476809

Here is the abstract from “FLIP: Cross-domain Face Anti-spoofing with Language Guidance.”

Face anti-spoofing (FAS) or presentation attack detection is an essential component of face recognition systems deployed in security-critical applications. Existing FAS methods have poor generalizability to unseen spoof types, camera sensors, and environmental conditions. Recently, vision transformer (ViT) models have been shown to be effective for the FAS task due to their ability to capture long-range dependencies among image patches. However, adaptive modules or auxiliary loss functions are often required to adapt pre-trained ViT weights learned on large-scale datasets such as ImageNet. In this work, we first show that initializing ViTs with multimodal (e.g., CLIP) pre-trained weights improves generalizability for the FAS task, which is in line with the zero-shot transfer capabilities of vision-language pre-trained (VLP) models. We then propose a novel approach for robust cross-domain FAS by grounding visual representations with the help of natural language. Specifically, we show that aligning the image representation with an ensemble of class descriptions (based on natural language semantics) improves FAS generalizability in low-data regimes. Finally, we propose a multimodal contrastive learning strategy to boost feature generalization further and bridge the gap between source and target domains. Extensive experiments on three standard protocols demonstrate that our method significantly outperforms the state-of-the-art methods, achieving better zero-shot transfer performance than five-shot transfer of “adaptive ViTs”.

From https://koushiksrivats.github.io/FLIP/?utm_source=tldrai

FLIP, by the way, stands for “Face Anti-Spoofing with Language-Image Pretraining.” CLIP is “contrastive language-image pre-training.”

While I knew I couldn’t master this, I did want to know what LIP and ViT were.

However, I couldn’t find something that just talked about LIP: all the sources I found talked about FLIP, CLIP, PLIP, GLIP, etc. So I gave up and looked at Matthew Brems’ easy-to-read explainer on CLIP:

CLIP is the first multimodal (in this case, vision and text) model tackling computer vision and was recently released by OpenAI on January 5, 2021….CLIP is a bridge between computer vision and natural language processing.

From https://www.kdnuggets.com/2021/03/beginners-guide-clip-model.html

Sadly, Brems didn’t address ViT, so I turned to Chinmay Bhalerao.

Vision Transformers work by first dividing the image into a sequence of patches. Each patch is then represented as a vector. The vectors for each patch are then fed into a Transformer encoder. The Transformer encoder is a stack of self-attention layers. Self-attention is a mechanism that allows the model to learn long-range dependencies between the patches. This is important for image classification, as it allows the model to learn how the different parts of an image contribute to its overall label.

The output of the Transformer encoder is a sequence of vectors. These vectors represent the features of the image. The features are then used to classify the image.

From https://medium.com/data-and-beyond/vision-transformers-vit-a-very-basic-introduction-6cd29a7e56f3

So Srivatsan et al combined tiny little bits of images with language representations to determine which images are (using my words) “fake fake fake.”

From https://www.youtube.com/shorts/7B9EiNHohHE

Because a bot can’t always recognize a mannequin.

Or perhaps the bot and the mannequin are in shenanigans.

The devil made them do it.