May 17, 2020
Over the last year, various Congressional committees have been investigating the expanding use of facial recognition technology by law enforcement and the private sector. In this episode, hear the highlights of these investigations which will enlighten you about the extent that this technology is being used to put your face in criminal investigation line-ups, determine your employability, and more.
Use your bank’s online bill pay function to mail contributions to:
Please make checks payable to Congressional Dish
Thank you for supporting truly independent media!
CD158: Rapid DNA Act
1:37:25 Rep. Lauren Underwood (IL): Some passengers report being unaware or confused about how to opt out of their biometric screening. As CBP expands the biometric screening program, does it intend to reevaluate the best method of communicating the important opt out information to passengers? John Wagner: Yeah, so right now we've got signage at the airports. But you know, a lot of people don't read signs at the airport. We've got gate announcements that the airlines try to make before boarding. But again, there's always competing announcements going on. And sometimes it's tough to understand what's being said. So we're actually looking with the airlines is - could we print things on the boarding pass could we give notifications when they're, say booking their ticket or when they're getting their their checking information for boarding other electronic messages we could provide, so we're looking at additional ways to do that. We also started taking out some some privacy advertisements, advising people of the requirements and what their options are as well, too.
2:40:00 Christopher Wray: We at the FBI don't use facial recognition for anything other than lead value. There is no one under FBI policy who is arrested, much less convicted based on facial recognition technology. We use it to advance an investigation to then be used with other information to figure out if we’re going in the right place. So let me start with that. Second thing. We scrupulously train all the examiners under various constitutional protections. And then as to the DMV searches that you're talking about, again we the FBI don't do those searches. The only way those searches can happen is under strict MOUs that have all kinds of constitutional backing. Even when we get the results, it then has to be reviewed carefully by a trained examiner.
2:41:00 Rep. Pramila Jayapal (WA): To be clear under, current FBI policy, can face recognition technology be used without a warrant or probable cause in any circumstance? Christopher Wray: Yes. Rep. Pramila Jayapal (WA): OK, so that is a concern for me. It continues to be a concern for me.
40:55 Charles Romine: I'll first address one-to-one verification applications. Their false positive differentials are much larger than those related to false negative and exist across many of the algorithms tested. False positives might present a security concern to the system owner as they may allow access to imposters. Other findings are that false positives are higher in women than in men and are higher in the elderly and the young compared to middle aged adults. Regarding race, we measured higher false positive rates in Asian and African American faces relative to those of Caucasians. There are also higher false positive rates in Native American, American Indian, Alaskan Indian and Pacific Islanders. These effects apply to most algorithms, including those developed in Europe and the United States. However, a notable exception was for some algorithms developed in Asian countries. There was no such dramatic difference in false positives in one to one matching between Asian and Caucasian faces for algorithms developed in Asia. This study did not explore the relationship between cause and effect, one possible connection and an area for research is the relationship between an algorithms performance and the data used to train the algorithm itself.
1:13:00 Meredith Whittaker: The average consumer does not and indeed many researchers, many lawmakers don't because this technology, as I wrote about my my written testimony, is hidden behind trade secrecy. This is a corporate technology that is not open for scrutiny and auditing by external experts. I think it's notable that while NIST reviewed 189 algorithms for their latest report, Amazon refused to submit their recognition algorithm to NIST. Now, they claimed they couldn't modify it to meet NIST standards, but they are a multi billion dollar company and have managed some other pretty incredible feats. So whatever the reason is, what we see here is that it's at the facial recognition companies discretion, what they do or don't release.
1:51:45 Meredith Whittaker: Because the Baltimore PD was using private sector technologies, they were scanning Instagram photos through a service called Geopedia that gave them feeds from Freddie Gray protests. They then were matching those photos against their Faces facial recognition algorithm which is a privately developed facial recognition algorithm to identify people with warrants, whom they could then potentially harass.
2:49:45 Rep. Deb Haaland (NM): I recently read that some employers have begun using facial recognition technology to help decide who to hire. At certain companies such as Hilton and Unilever, job applicants can complete video interviews using their computer or cell phone cameras which collect data on characteristics like an applicant's facial movements, vocal tone and word choice. One company offering this technology, HireVue, collects up to 500,000 data points in a 30 minute interview. The algorithm then ranks the applicant against other applicants based on the so called employability score. Job applicants who look and sound like the most like the current high performers at the company received the highest scores. Miss Whittaker, I have two questions for you. One, isn't it true that the use of facial recognition and characterization technology and job application processes may contribute to biases in hiring practices. And if yes, can you please elaborate? Meredith Whittaker: It is absolutely true. And if the scenario that you described so well is a scenario in which you create a bias feedback loop, in which the people who are already rewarded and promoted and hired to a firm become the models for what a good employee looks like. So if you look at the executive suite at Goldman Sachs, which also uses HireVue, for this type of hiring, you see a lot of men, a lot of white men, and if that becomes the model for what a successful worker looks like, and then that is that is used to judge whether my face looks successful enough to get a job interview at Goldman Sachs, we're going to see a kind of confirmation bias in which people are excluded from opportunity because they happen not to look like the people who had already been hired.
2:54:45 Rep. Jim Jordan (OH): First part of what we hope will be legislation that we can have broad support on, that the chairman and both Republicans and Democrats can support, is tell us what's going on now. And then second, while we're trying to figure that out, while the study and we're getting an accountability and what's all happening, let's not expand it. Let's just start there, tell us what you're doing, and don't do anything while we're trying to figure out what you're doing. And then once we get that information, then we can move from there. That is what I hope we can start with Madam Chair and frankly, what we've been working with now for a year, the staffs for both majority and the minority.
4:55 Rep. Bennie Thompson (MS): Last July, the American Civil Liberties Union connected..conducted a test using Amazon's facial recognition to call recognition. ACLU built a database of 25,000 publicly available arrest photos. Using recognition, ACLU searched the database using pictures of every current member of Congress. That software incorrectly matched 20 members, 28 members with individuals who had criminal records.
10:30 Rep. Mike Rogers (AL): I do not believe that anyone has a reasonable expectation of privacy in a government ID photo. Period.
15:00 John Wagner: CBP developed a service that simply automates the manual facial recognition process that goes on today when a traveler presents a passport to establish their identity. To be clear, CBP is only comparing the picture taken against photos previously provided by travelers to the U.S. Government for the purposes of international travel. This is not a surveillance program.
16:10 John Wagner: Now recognizing there's been concerns raised over the inclusion of US citizens, CBP has existing authorities and responsibilities to determine the citizenship and identity of all people traveling internationally. This is a U.S. government responsibility, not the private sector. It's also unlawful for a U.S. Citizen to travel internationally without a U.S. Passport. Now, generally determination of U.S. Citizenship is done by comparing the traveler against their passport. Again, we're simply automating and using a computer algorithm to enhance this manual facial recognition existing process.
16:50 John Wagner: As far as our partnerships with the industry stakeholders, CBP'S developed a standard set of business requirements that our partners have all agreed too, If their camera is sending a photo to CBP. The business requirements clearly stipulate they cannot keep the photos. Our partners have voluntarily agreed to the CBP business requirements.
25:40 Joseph R. DiPietro: With respect to DNA, DNA evidence is one of the most effective identification tools available to law enforcement today. Advances related to DNA technology have been rapid and the secret service remains dedicated to utilizing new applications to enhance our integrated mission.
26:55 Joseph R. DiPietro: The secret service is currently working on a facial recognition pilot. The participants in the pilot are secret service employees who volunteer to take part in this effort. Designated White House cameras that are part of the video management system captured volunteers as they move through various locations around the White House complex. Software running on a server,dedicated to the pilot, and on a closed network not connected to the Internet, seeks to match the images of the volunteers to the images in the video streams.
37:40 John Wagner: So when the picture is taken and provided and comes into CBP and we match it against one of our pre-staged gallery photos that's comprised of passports and visas and previous arrivals, if it's a foreign national subject to the biometric entry exit mandate, that photograph will be sent over to DHS to hold them-to be stored in IDENT, which is the departments repository for that information. If it's a U.S. citizen and that document-that photo matches a U.S. Passport or a permanent resident or somebody outside of the scope of entry exit, that photograph would be held for 12 hours and then deleted or purged from our systems. The only reason we hold it for that short period of time, is just in case the system crashes and we have to restore everything.
38:45 John Wagner: What we were doing with that subcontractors, we would testing their camera on the U.S. Mexico land border in a standalone pilot system. So it wasn't integrated into the main CBP network and we were testing the taking of the photographs and the license plates and the ability to take a picture of a person in a vehicle and whether that would be matchable. In this case, the, apparently the con-...as far as I understand, the contractor physically removed those photographs from the camera itself and put it onto their own network, which was then breached. The CBP network was not hacked. The contractor, and what we see is, what I believe is, they remove that in violation of the contract and that's why our relationship has been severed with them and we're conducting an investigation. Rep. Bennie Thompson (MS): So - so you see my concern about how we control the data we collect? John Wagner: Absolutely.
1:08:00 Austin Gould: So right now I can comment on a really what we're doing in Atlanta with Delta Airlines. In Atlanta, the Delta airlines kiosks use biometric identification to-, when the passenger checks in to make sure, should, they choose to do, too make sure that that person is actually the passenger who's ticketed on that particular flight. Uh, TSA has oversight of the bagdrop to ensure that passengers are positively matched to bags in the international, you know, for international travel. And so Delta Airlines has a security program amendment that we've granted them to use biometric technology to do that matching at the bagdrop. We use it at our checkpoint in uh, in Atlanta, and then it's a, of course subject, or it's used at the exit point at the gate.
1:08:55 Austin Gould: Right now, the Security Program amendment that we've granted Delta for the limited use, only in Atlanta, is the only formal agreement that we've entered into with the, uh, with the airlines.
1:20:45 Rep. Yvette Clarke (NY): I'm concerned about the lack of accuracy. I'm very concerned about.... John Wagner: A person doesn't match the photo in this case, they present their passport as they're doing today. Rep. Yvette Clarke (NY): Excuse me? John Wagner: If a person doesn't match a photograph, they simply present their passport... Rep. Yvette Clarke (NY): When you're trying to match them and they don't match what happens to that individual? John Wagner: They present their boarding pass and their passport... Rep. Yvette Clarke (NY): Uh huh. John Wagner: ...and it's manually reviewed at that point in time. Just as it happens today. Rep. Yvette Clarke (NY): is that, and those people aren't detained in any way? They're not asked to step aside, they're not asked to, the process does not delay that person? John Wagner: No, they just show their passport. Rep. Yvette Clarke (NY): Okay. I hope that's the case.
1:33:00 Joseph R. DiPietro: Ma'am, the cameras that we're using as part of this pilot are part of the White House video management system. That's the CCTV system that records videos from all the cameras around the complex. We retain that data for 30 days as part of the CCTV process. So if we're, as we're going through and we're identifying those, those volunteers that are in there, that record is saved and we save that and we're going to evaluate that until the end of the process.
1:36:30 Rep. Debbie Lesko (AZ): Mr. Gould, are you planning on using this or have you thought of using biometric technology or do you for the employees-, the airport employees? Austin Gould: Yes ma'am. We are considering using biometric identification processes for employees as well.
1:42:00 John Wagner: This is not us taking an image of a person and randomly running it against a gallery set of indistinguishable, say, quality photographs and lowering down the accuracy rate as to what constitutes a match, to make it match someone that it's not.
50:00 David Pekoske: Right now, based on a series of rules, a passenger who was not a precheck register or a global entry registrant could get precheck on their boarding pass. We're phasing that out over the course of the next several months. Um, so the precheck experience should get quite a bit better.
2:30 Kimberly Del Greco: The FBI's policy and procedures emphasize that photo candidates returned are not to be considered positive identification, that the searches are photos and will only result in a ranked listing of candidates.
3:15 Kimberly Del Greco: Photos in the NGI, IPS repository are solely criminal mugshots acquired by law enforcement partners with criminal fingerprints associated with an arrest.
3:25 Kimberly Del Greco: The FBI face services unit provides investigative lead support to FBI offices, operational divisions, and legal attache's by using trained face examiners to compare face images of persons associated with open assessments or active investigations against facial images available and state and federal facial recognition systems through establish agreements with state and federal authorities.
3:50Kimberly Del Greco: The face services unit only searches probe photos that have been collected pursuant to the attorney general guidelines as part of an authorized FBI investigation and they are not retained.
4:05 Kimberly Del Greco: This service does not provide positive identification, but rather an investigative lead.
4:45 Kimberly Del Greco: The FBI collaborated with NIS to perform the facial recognition vendor test and determined a most viable option to upgrade its current NGI IPS algorithm. The algorithm chosen boasted an accuracy rate of 99.12% leveraging the nest results. The FBI is implementing the upgraded facial recognition algorithm.
7:30 Gretta Goodwin: We also reported on accuracy concerns about FBI's face recognition capabilities. Specifically, we found that the FBI conducted limited assessments of the accuracy of the face recognition searches before they accept it and deployed the technology. For example, the face recognition system generates a list of the requested number of photos. The FBI only assessed accuracy when users requested a list of 50 possible matches. It did not test smaller list sizes, which might have yielded different results. Additionally, these tests did not specify how often incorrect matches were returned. Knowing all of this, the FBI still deployed the technology.
13:30 Charles Romine: NIST's face recognition vendor testing program was established in 2000 to provide independent evaluations of both prototype and commercially available facial recognition algorithms. Significant progress has been made in algorithm improvements since the program was created.
14:30 Charles Romine Optimal face identification was achieved only when humans and machines collaborated.
16:40 Austin Gould The roadmap has four major goals, partnered with customs and border protection on biometrics for international travelers, operationalize biometrics for TSA precheck passengers, potentially expand biometrics for additional domestic travelers and develop the infrastructure to support these biometric efforts.
17:00 Austin Gould Consistent with the biometrics roadmap, TSA has conducted pilots that use facial biometrics to verify identity at certain airports.
17:25 Austin Gould And passengers always have the opportunity to not participate. In these cases, standard manual identification process is used.
17:30 Austin Gould I have observed the pilot currently underway in Terminal F in Atlanta for international passengers. Of Note, virtually every passenger chose to use the biometric identification process. The facial capture camera used for this pilot was in active mode, meaning that it only captured a facial image after the passenger was in position and the officer activated it. The match rate is extremely high and passengers moved rapidly through the checkpoint.
20:45 Rep. Elijah Cummings (MD) Ms. DelGreco, can you explain how the FBI decides to search a state database versus when it searches its own system and how this policy is determined? Kimberly Del Greco I'd be happy to explain that. The, at the FBI, we have a service called Face Services unit. They process background checks and, uh, process, facial recognition searches of the state DMV photos. They do this in accordance with the attorney general guidelines. An FBI field office has to have an open assessment or an active investigation. They submit the probe photo to the FBI Face Services unit. We launched the search to the state. The state runs the search for the FBI and, and provides a candidate list back.
21:35 Kimberly Del Greco With regard to the NGI IPS, the Interstate Photo system, the Face Services unit will utilize that repository as well as the DMV photos. However, state and local and federal law enforcement agencies only have access to the NGI Interstate Photo system. These are the FBI mugshots that are associated with an 10 print criminal card associated with a criminal arrest record.
22:05 Rep. Elijah Cummings (MD) Well, do individual who consent to having their faces in the noncriminal databases also consent to having their faces searched by the FBI for criminal investigations? For example, when applying for a drivers license, does someone consent at the DMV to being in a database searchable by the FBI? Kimberly Del Greco The FBI worked diligently with the state representatives in each of the states that we have MOUs. We did so under the states’ authority to allow photos to be used for criminal investigations. We also abided by the Federal Drivers License Privacy Protection Act and we consider that a very important process for us to access those photos to assist the state and local law enforcement and our Federal agencies. Rep. Elijah Cummings (MD) Well, you just said state authority allowed you to do this. One question that our ranking member has asked over and over again is do you know whether in any of these states do any elected officials have anything to do with these decisions? In other words, where is that authority coming from and we’re trying to figure out if something affecting so many citizens whether elected officials have anything to do with it. Do you know? Kimberly Del Greco I do. Only in one state - the state of Illinois - did an elected official sign the MOU. In the other states, they were done so with the state representatives. This is state law that’s established at the state level prior to facial recognition and our program getting started. We’re just leveraging that state law. That state law is already in place. We did work with the office of general council at the FBI and the attorney level at the state level. Rep. Elijah Cummings (MD) Well, if it was prior to facial recognition coming into existence, I’m just wondering do you think that whatever laws you’re referring to anticipated something like facial recognition? Kimberly Del Greco It’s my understanding that the states established those laws because of fraud and abuse of drivers licenses and we are just reviewing each of the state laws and working with the representatives in those states to ensure that we can leverage that for criminal investigation. Rep. Elijah Cummings (MD) And so when you say leverage, I guess you’re saying that there were laws that were out there and these laws did not anticipate something like facial recognition and now the FBI has decided that it would basically take advantage of those laws, is that right?
26:00 Rep. Elijah Cummings (MD) Ms. Del Greco, how many states have provided this level of direct access to the FBI? Kimberly Del Greco We do not have direct access. We submit a probe to the state. There’s 21 states… Rep. Elijah Cummings (MD) 21 states, ok.
28:10 Rep. Elijah Cummings (MD) And can the FBI perform a face recognition service for any American with a passport? Kimberly Del Greco For an open assessment or an active investigation. Only by the FBI, sir.
29:25 Kimberly Del Greco Some of those successes are assisting with the capture of the terrorist in Boston.
31:15 Rep. Paul Gosar (AZ) So what sort of accuracy rates are you finding in the different algorithms ability to match an image against a larger gallery of images? Charles Romine The accuracy rates that we're seeing, we have many different participants who have submitted algorithms. Approximately 70 participants in our, in our testing, the best algorithms are performing at a rate of approximately 99.7 in terms of accuracy. There's still a wide variety or wide variance across the number of algorithms. So this is certainly not commoditized yet. Some of the participants faired significantly poorer than that.
32:00 Rep. Paul Gosar (AZ)So are there algorithms that you tested that you would recommend for law enforcement? Charles Romine We don't make recommendations about specific algorithms. We provide the data necessary for making informed decisions about how an algorithm will perform in a field.
32:20 Charles Romine For law enforcement, for example, accuracy rates are one important aspect that needs to be considered, but there are other aspects that have to be taken into consideration for procurement or acquisition of such.
34:15 Rep. Stephen Lynch (MA): Mr.
Gould, according to the biometrics roadmap released by TSA in
September of 2018, TSA seeks to expand the use of facial
recognition technology to "the general flying public" in specific
locations. But the "general flying public "and TSA envisions the
use of technology upon domestic flights, as well as international,
which would capture the faces of mostly American citizens, and I'm
just curious, going back to the chairman's original question,
what's the legal basis? I'm not talking about a situation with the
FBI where you might have, you hopefully would have probable cause.
Where does the TSA find its justification? Its legal justification
for capturing the facial, uh, identity of, of the flying public.
Austin Gould: Yes sir. In accordance with the
Aviation Transportation Security Act of 2001, TSA is charged with
positively identifying passengers who are boarding aircraft. That
probably... Rep. Stephen Lynch (MA): Right. Let me
just stop you right there. So, we all fly at least a couple of
times a day.... Austin Gould: Yes sir.
Rep. Stephen Lynch (MA) : ..a week. So we have, now you have to have a certified license. You can't go with the old version that your State had. Now we have much more accurate licenses. We surrender that oftentimes in the airport during the boarding process, you've got to show it a couple of times you've got a ticketing issue there. So you're doing that right now. Austin Gould: Yes Sir. Rep. Stephen Lynch (MA) : You have been doing that for a long, long time. Austin Gould : Manually, Yes Sir. Rep. Stephen Lynch (MA) : Right. Right. Right. So now you're saying that you're going to do these pilot programs and you're gonna hurt people. Now you're saying voluntarily, but I could imagine like you've done with a pre-check, you can either agree to surrender your right to anonymity and wait in the long line or you can give up your fourth amendment rights and go in the quick line. Is is that the dynamic that's going on here? Austin Gould: Sir, with respect to expanding to the general traveling public, we anticipate using, and we've not tested this yet, a one to one matching capability at the checkpoint. You produce your credential, you stick it in a machine, and the machine identifies whether or not your image, which is captured by the camera, matches the image that's embedded in the credential and it returns a match result. That will then allow you to proceed through the checkpoint. Should you decide not to participate in that program, we will always have the option to do that process manually. Rep. Stephen Lynch (MA) : Right, but to match, you've got to have that data in the.. you've got to have that data onboarding and the technology to begin with to match something with, right? Austin Gould : Sir, that data is embedded in your credential. So the photograph is on your driver's license, for example. There's a digital recording of that image in the credential and when your pictures captured by the camera, it is matched to the photograph on the credential. It does not depart the checkpoint for any database search or anything like that. Rep. Stephen Lynch (MA) : Okay. Austin Gould : That's the one to one identification that we intend to use for the broader traveling public.
37:34 Rep. Stephen Lynch (MA) : You don't anticipate taking, using a database or gathering, collecting a database of information with NTSA, with which to identify passengers? Austin Gould : Sir, for international travelers who have a passport photo on record and for TSA precheck passengers who also provide a passport photo, we will match them to a gallery. But for the general traveling public that does not participate in those programs and merely has a credential, that matching.... Rep. Stephen Lynch (MA) : What's the size of the gallery,? What do you anticipate? Is that, so if anybody engages in international travel, Is that, are they going to be in that or are they foreign nationals who traveled to the U.S.? Austin Gould : Sir, the gallery that we use right now with TVS includes anyone who is traveling internationally and who has a photo on record.
49:40 Rep. William Lacy Clay (MO) : So how many times has the FBI provided notice to criminal defendants that face recognition was used in their case? Kimberly Del Greco : As part of a criminal investigation, I don't believe that's part of the process.
52:00 Rep. Jim Jordan (OH) : Dr. Goodwin, did the FBI publish privacy impact assessment in a timely fashion as it was supposed to when it implemented FRT in 2011? Gretta Goodwin : No. Rep. Jim Jordan (OH) : Did the FBI follow proper notice? File proper notice, specifically the system of record notice in a timely fashion when it implemented facial recognition technology? Gretta Goodwin : No. Rep. Jim Jordan (OH) : Did the FBI conduct proper testing of the next generation interstate photo system when it implemented FRT? Gretta Goodwin : Proper in terms of its accuracy for its use? Rep. Jim Jordan (OH) : Yes Gretta Goodwin : No. Rep. Jim Jordan (OH) : Did the FBI test the accuracy of the states systems that it interfaced with? Gretta Goodwin : No.
58:00 Rep. Carolyn Maloney (NY) : So Mrs. Del Greco, how many searches has the FBI run in the Next Generation ID Interstate Photo system to date? How many searches? Do you have that information? Kimberly Del Greco : I have, I have from fiscal year 2017 to April of 2019. There were 152,500 searches.
1:03:30 Rep. Thomas Massie (KY) : Did you test certain conditions like siblings, the accuracy for siblings? Charles Romine : We do have perhaps the most relevant data that I can give you is, we do know that there is an impact on twins, in the database or in the testing, whether they are identical twins or even fraternal twins.
1:14:50 Austin Gould : Sir, the system that TSA is prototyping in conjunction with CBP uses NEC camera and a matching algorithm that was also developed by NEC.
1:16:50 Rep. Justin Amash (MI) : Do you have plans to implement face recognition technology at additional points in airports beyond besides gates or security checkpoints? Austin Gould : We are prototyping facial recognition technology at bagdrops, so when you drop a bag off to be placed on an aircraft, we can use facial technology, we're exploring the use of facial technology there and then for TSA purposes, only other locations are the checkpoint.
1:17:20 Rep. Mark Meadows (NC) : So
Mr. Gould, let me, let me come back. If you're doing at bagdrops,
that's not a one on one comparison. I mean if you, what are you
comparing it to? If you're, if you're looking at change, checking
facial recognition at bagdrops... Austin Gould :
Rep. Mark Meadows (NC) : ..there wouldn't be necessarily the identification that you were talking about earlier. What pilot program are you working with that? Austin Gould : The pilot program in place right now is with Delta Airlines and CBP and TSA and Atlanta's Terminal F and it's a matching of the passengers bag against their identification or their photograph and the TVS, CBP, TVS system. Rep. Mark Meadows (NC) : Well, that contradicts your earlier testimony, Mr. Gould. Because what you said that you were doing is just checking the biometrics within the identification against a facial recognition, but it sounds like you're doing a lot more than that.
1:18:50 Austin Gould: Sir, with respect to the pilot in Atlanta, it's international travelers, and the purpose of that pilot is to positively match using biometrics. The passenger to that bag at the bag drop. The only, the traveler's camp, uh, photograph is captured, images captured. It is transmitted to the CBP TVS system for matching and it returns a match result. That's it. No privacy information or any other data associated with it.
1:41:30 Rep. Jim Jordan (OH): The numbers. Dr. Goodwin, how many..what number of photos does the FBI have access to in just their database? Gretta Goodwin: In just their database, it's a little over 20 plus, 36 million. Rep. Jim Jordan (OH): 36 million. And then in the databases that they can then send information to and that are screened and used and there's interface, interaction with, at the state level. What is the total number of photos in those databases? Gretta Goodwin: So access to photos across all the repositories? About 640 million. Rep. Jim Jordan (OH): 640 million photos? Only 330 million people in the country.
1:45:35 Charles Romine: We don't test for specific companies on their behalf. We test or evaluate the algorithms that are submitted to us through this voluntary program.
1:45:45 Charles Romine: We don't test specifically for Algorithms, demographic effects. We're talking about the demographic effects across all of the Algorithms that are submitted.
1:49:30 Rep. Mark Meadows (NC): Is you mentioned about not having any real time systems, and yet we had a testimony just a couple of weeks ago from Georgetown that indicated that Chicago Police Department, Detroit Police Department has real-time. They purchased it where they're actually taking real-time images. Do They Ping the FBI to validate what they've picked up in real-time with what you have on your database? Kimberly Del Greco: I mean, there are authorized law enforcement entities that have access to our system.
1:53:25 Rep. Mark Meadows (NC): I would suggest that you put this pilot program on hold, because I don't know of any appropriations that specifically allowed you to have this, this pilot program. Are you aware of any? Because you keep referring back to a 2001 law, and I'm not, I'm not aware of any appropriations that have been given you the right to do this pilot program. Austin Gould: I'm not aware of any specific appropriations. Rep. Mark Meadows (NC): Exactly, so I would recommend that you stop it until you find out your statutory authority.
2:29:12 Rep. Alexandria Ocasio-Cortez (NY): The TSA has outlined proposals to collaborate with private companies including Delta and Jet Bue to develop and implement their facial recognition search systems. Is this correct? Austin Gould: Ma'am, we've issued a security program amendment to Delta to allow them to use biometric identification at their bagdrop. In terms of partnering with them to develop the backend matching system, that is something that we're solely engaged withCBP on..... Rep. Alexandria Ocasio-Cortez (NY): And the bagdrop, those are the computers that folks check in and get their boarding pass from? Austin Gould: That would be the, I would use the term "kiosk" for that. Rep. Alexandria Ocasio-Cortez (NY): "The kiosk?" Austin Gould: Delta uses that technology at their kiosk. TSA has no equity there, that's solely to verify that passenger has a reservation with Delta where we have equities that are checkpoint and also at the bagdrop where we're required to ensure that the passengers match to their bag. Rep. Alexandria Ocasio-Cortez (NY): Do individuals know that that is happening and do they provide explicit consent? Is it opt in? Austin Gould: Passengers have the opportunity to not participate. Rep. Alexandria Ocasio-Cortez (NY): So it's opt out, but not opt in? Austin Gould: It is, yes ma'am. Rep. Alexandria Ocasio-Cortez (NY): So it's possible that jet blue and Delta are working with the TSA to capture photos of passengers faces without their explicit opt-in consent? Austin Gould: Man, I was down in Atlanta last week and watched the Delta Check-in Process, the bagdrop process, and it was very clear, while I was down there, the passengers were afforded the opportunity, if you'd like to use, you know, facial capture for identification, please stand in front of the camera and we'll do so. There was no automatic capture of passengers or anything like that. Rep. Alexandria Ocasio-Cortez (NY): And this capture is not saved in any way, but is a..-correct, right? Austin Gould: No, ma'am. The camera captures the image. The image is encrypted. It is sent to the TVS matching system, which is what CBP uses solely for the purpose of match. And then that match result is sent back to to the operator. Rep. Alexandria Ocasio-Cortez (NY): Is that captured image destroyed? Austin Gould: It's not retained at all. No, ma'am. Rep. Alexandria Ocasio-Cortez (NY): So it's sent, but it's not retained? Austin Gould: It's not retained on the camera. No, ma'am.
13:15 Joy Buolamwini Due to the consequences of failures of this technology, I decided to focus my MIT research on the accuracy of facial analysis systems. These studies found that for the task of guessing a gender of a face; IBM, Microsoft, and Amazon had errors of no more than 1% for lighter-skinned men. In the worst case, those errors rose to over 30% for darker skinned women. Given such accuracy disparities, I wondered how large tech companies could have missed these issues. It boiled down to problematic dataset choices. In evaluating benchmark data sets from organizations like NIST, (The National Institute for Standards and technology) I found some surprising imbalances. One missed dataset was 75% male and 80% lighter skin, or what I like to call a pale male dataset. We cannot adequately evaluate facial analysis technologies without addressing this critical issue. Moving forward, the demographic and phenotypic composition of missed benchmarks must be made public and updated to better inform decision makers about the maturity of facial analysis technology.
21:30 Clare Garvie Face recognition gives law enforcement a power that they've never had before and this power raises questions about our fourth and first amendment protections. Police can't secretly fingerprint a crowd of people from across the street. They also can't walk through that crowd demanding that everybody produce their driver's license, but they can scan their faces,remotely and in secret, and identify each person thanks to face recognition technology.
22:00 Clare Garvie Last year, the Supreme Court in Carpenter noted that for the government to secretly monitor and catalog every one of our movements across time and space violates our right to privacy protected by the fourth amendment. Face recognition enables precisely this type of monitoring, but that hasn't stopped Chicago, Detroit, and other cities from acquiring and piloting this capability. The Supreme Court held in NAACP vs. Alabama, Tally vs. California, and others that the first amendment protects the right to anonymous speech and association. Face recognition technology threatens to upend this protection.
23:00 Clare Garvie Face recognition makes mistakes and its consequences will be born disproportionately by African Americans. 1. Comunities of color are disproportionately the targets of police surveillance, face recognition being no exception. San Diego found that their police used face recognition up to a two and a half times more on African Americans than on anyone else. 2. People of color are disproportionately enrolled in police face recognition systems, thanks to being over-represented in mugshot databases that the systems run on. And 3, Studies continue to show that the accuracy of face recognition varies depending on the race of the person being searched. Face recognition makes mistakes and risks making more mistakes, more misidentification's of African Americans. And the state could mean you're accused of a crime you didn't commit, like the Brown University student erroneously identified as one of the Sri Lankan bombers earlier this month. One of this country's foundational principles is equal protection under the law. Police use of face recognition may not comport with this principle.
24:05 Clare Garvie Left unchecked, current police face recognition practices threaten our due process rights. My research has uncovered the fact that police submit what can only be described as garbage data into face recognition systems expecting valuable leads in return. The NYPD submitted a photo of actor Woody Harrelson to find an unknown suspect in a beer theft. They have submitted photos of suspect whose eyes are mouths have been cut and pasted in from another person's photo, essentially, fabricating evidence. Agencies submit drawings of suspects in places of photos as well, despite research showing that this will not work. Worse, officers' at times then skip identification procedures and go straight to arresting someone on the basis of a face recognition search. This practice runs counter both to common sense and to department's own policies and these practices raised serious concerns about accuracy and the innocence of the person arrested because of a face recognition search.
25:15 Clare Garvie These systems produce Brady material, information that under our constitutional right to due process must be turned over to the defense, but it's not.
25:25 Clare Garvie For all these reasons, a moratorium on the use of face recognition by police is both appropriate and necessary.
30:15 Neema Singh Guliani The committee should Look at companies that are aggressively marketing this technology to the government, including how accurate their technologies are and what responsibility they take to prevent abuse. Companies are marketing this technology for serious uses, like identifying someone during a police encounter, and we know far too little. For example, Amazon has even refused to disclose who it sells this technology too and companies like Microsoft and Face Burst have so far not received significant congressional attention.
30:45 Neema Singh Guliani There are efforts across the country to stop this dangerous spread of this technology. San Francisco has banned the use by city departments and Amazon shareholders are today taking the unprecedented step of voting on a resolution that would stop the company from selling this technology to the government and force it to study the human rights impacts. Congress should follow these good examples and put in place a moratorium on law enforcement use.
39:44 Rep. Katie Hill (CA) Professor Ferguson, do you think that the supreme court can rule quickly enough upon the use of these technologies as the cases arise to thwart constitutionally questionable uses? Andrew Ferguson They can, but they won't do as good a job as congress regulating it. Now, Justice Alito has repeatedly made that claim, and I think he's correct to say that this kind of technology should be regulated first, by Congress. The fourth amendment floor will exist and the Supreme Court will address it. But this body has the primary responsibility to regulate in this field.
44:57 Rep. Jim Jordan (OH) Did the state legislature and the governor actually pass legislation saying it was okay for the FBI to access every single person in their state who has a driver's license? Did that happen in those 18 or 19 states that gave that permission to the FBI? Neema Singh Guliani No, and that's the problem. This was all in secret essentially. Rep. Jim Jordan (OH) So some unelected person at the FBI talks to some unelected person at the state level and they say, "yeah, go ahead". Here's.. In the case of Ohio, we've got 11 million people, most of them drive, here's 10 million folks who you can now have their, have this database. Neema Singh Guliani Right, and the people who wanted a driver's license many times didn't know these systems were operating either.
1:02:16 Rep. Michael Cloud (TX) Miss Buolamwini, did I say that right? Joy Buolamwini Yes, You did. Rep. Michael Cloud (TX) Okay. You mentioned Facebook, in your remarks and I find that interesting cause I'm extremely concerned about the government having this kind of unknown checked ability. I would be curious to get your thoughts of corporations having the same sort of ability and also Ms.Garvie and Ms. Guliani, if you want to speak to that too. Joy Buolamwini Absolutely. So you're looking at a platform that has over 2.6 billion users and over time, Facebook has been able to amass enormous facial recognition capabilities using all of those photos that we tagged without our permission. What we're seeing is that we don't necessarily have to accept this as the default. So in the EU where GDPR was passed, because there's a provision for biometric data consent, they actually have an option where you have to opt in. Right now we don't have that in the US and that's something we could immediately require today.
1:09:10 Joy Buolamwini We don't even have reporting requirements, at least in the UK where they have done pilots of facial recognition technology. There are reported results and you have false positive match rates of over 90%. There's a big brother Watch UK report that came out that showed more than 2,400 innocent people had their faces misidentified.
1:13:05 Clare Garvie Law enforcement agencies don't typically have access to the training data or to how the algorithms work as well, because these are private companies that have developed these systems and it's considered a trade secret.
1:14:22 Clare Garvie We see China as a bit of a roadmap of what's possible with this technology in the absence of rules. And in the absence of rules, this is a system where everybody is enrolled in the backend and there are enough cameras to allow law enforcement to track where somebody is anytime they show their face in public, to upload their photo and see where they've been over the last two weeks, be that public rallies or an alcoholics anonymous meeting or, a rehab clinic. That information is now available at the click of a button or the upload of a photo. That's what face recognition looks like with no rules.
1:15:14 Clare Garvie Our research has found that, two, at least two major jurisdictions, Chicago and Detroit have purchased this capability and have paid to keep it, to maintain it. Chicago says they do not use it. Detroit, did not deny that they were using it. There's is designed to operate with project greenlight, which is specifically locations like, yes, gas stations and liquor stores, but also churches and clinics and schools.
1:41:41 Clare Garvie A handful of other agencies across the country, Los Angeles, the West Virginia Intelligence Fusion Center, and others have either piloted or have looked to purchase this technology as well.
1:41:55 Rep. Carol Miller (WV) Are there any federal agencies to your knowledge that utilize real time face surveillance? Clare Garvie The U.S. Secret service is piloting a program around the White House complex as we speak. We do not know the degree to which the FBI has been piloting this. We do know they have acquired or have been using Amazon recognition, which is the same, uh, surveillance capability that Orlando has been piloting in real time. But there is no transparency into how an when they're using that.
1:44:55 Rep. Alexandria Ocasio-Cortez (NY) Ms. Buolamwini, right now, Amazon can scan your face without your consent, all of our faces without our consent and sell it to the government, all without our knowledge, correct? Joy Buolamwini Yes. Rep. Alexandria Ocasio-Cortez (NY) And you know, Mr Chair, I'd like to seek unanimous consent on how Amazon actually met with ICE officials over facial recognition systems that could identify immigrants. I'd like to submit this to the congressional record. Chairperson Without objection. Rep. Alexandria Ocasio-Cortez (NY) Thank you so much. Um, Miss Garvie, in fact, it's not just Amazon that's doing this right? It's Facebook. It's Microsoft. It's a very large amount of tech corporations, correct? Clare Garvie That's correct. Rep. Alexandria Ocasio-Cortez (NY) And you think it's fair to say that Americans are essentially being spied on and surveilled on a massive scale without their consent or knowledge? Clare Garvie I would make a bit of a distinction between what Facebook, and other companies are doing, but yielding to Miss Buolamwini for more specifics on this. I will say most of the law enforcement agency systems operate on DMV databases or mugshot databases, so information that has been collected by agencies rather than companies.
1:50:15 Joy Buolamwini So there's a case with Mr. Bah, an 18 year old African American man who was misidentified in Apple stores as a thief. And in fact, he was falsely arrested multiple times because of this kind of misidentification.
2:07:50 Rep. Jimmy Gomez (CA) Until February of this year, Amazon had not submitted its controversial facial recognition technology recognition to third party testing with the National Institute of Standards and technology known as NIST. In a January, 2019 blog post, Amazon stated that "Amazon recognition can't be downloaded for testing outside of Amazon." In short, Amazon would not submit to outside testing of their algorithm. Despite the fact that Amazon had not submitted its facial recognition product to outside testing, it still sold that product to police departments. In 2017, police in Washington county, Oregon started using Amazon recognition technology.
2:28:15 Rep. Gerald Connolly (VA) The ubiquity of this technology, it strikes me, maybe we've already kind of mostly lost this battle.
2:36:30 Rep. Jamie Raskin (MD) We are now seeing that most companies that develop facial recognition systems offer also real time software. Do we know how many of these are selling their technology to government actors in the United States? Clare Garvie That's right. Most, if not all companies that market face recognition to law enforcement in the U.S., also advertise the abilities to do face surveillance. We have no idea how widespread this is thanks to a fundamental absence of transparency. We have limited visibility into what Chicago is doing, what Detroit's doing. Orlando, the secret service here in Washington, D.C. and in New York, thanks to FOIA records and investigative journalists work. But for a vast majority of jurisdictions, we have no idea.
2:37:20 Rep. Jamie Raskin (MD) But well, what's the minimum you think? Clare Garvie So, we can estimate conservatively that face recognition generally both used as an investigative tool, and potentially as a surveillance tool is, accessible to at very least, a quarter of all law enforcement agencies across the U.S. That's a conservative estimate because it's based on 300 or so records requests where there are 18,000 law enforcement agencies across the country.
2:39:00 Joy Buolamwini So Facebook has a patent, where they say because we have all of these space prints collected often without consent, we can now give you an option as a retailer to identify somebody who walks into the store and in their patent they say, "we can also give that face a trustworthiness score and based on that trustworthiness score, we might determine if you have access or not to a valuable good". So this... Rep. Jamie Raskin (MD) Facebook is selling this now? Joy Buolamwini This is a patent that they filed; as in something that they could potentially do with the capabilities they have, so as we're talking about state surveillance, we absolutely have to be thinking about corporate surveillance as well.
Design by Only Child Imaginations
Intro & Exit: Tired of Being Lied To by David Ippolito (found on Music Alley by mevio)