Secret Service trials facial recognition system around White House complex
You’d think the entirety of the “White House Complex,” as it is known inside the “Beltway,” would certainly by now have a biometric facial recognition system in place. But apparently it doesn’t. So, the US Secret Service (USSS) in November began the process of testing deployment of a Facial Recognition Pilot (FRP) program utilizing the existing Crown Closed Circuit Television (CCTV) camera system at the White House grounds “in order to biometrically confirm the identity of volunteer USSS employees in public spaces around the complex,” and, “to test USSS’s ability to verify the identities of a test population of volunteer USSS employees.”
That’s according to the November 26 Privacy Impact Assessment (PIA) issued by the USSS Office of Technical Development & Mission Support, and the Department of Homeland Security’s (DHS) Chief Privacy Officer.
Presently, the Secret Service uses Uniformed Division Officers and Special Agents consigned to the White House to identify subjects of interest that may approach the White House grounds.
“Ultimately, the goal of the FRP is to identify if facial recognition technologies can be of assistance to the USSS in identifying known subjects of interest prior to initial contact with law enforcement at the White House Complex,” the PIA explained, saying that the collection of volunteer subject data will support the Secret Service to test the proficiency of facial recognition technology “to identify known individuals, and to determine if biometric technology can be incorporated into the continuously evolving security plan at the White House Complex.”
This PIA addresses only Phase One of the FRP, which began November 19, and is expected to continue until August 30, 2019, at which time all data collected by the FRP will be deleted, unless, however, “it is associated with an open law enforcement matter.” The Secret Service will also “update this PIA, as necessary, to address any future field tests as they are developed and deployed.”
The Secret Service will take “precautions to prevent the alteration or deletion of the facial image data to ensure that all information is accurately captured and retained,” and, the “data will be stored on a USSS-approved server which will only be accessible by authorized USSS users. The data is for official use only, and is prohibited from being downloaded, manipulated, or otherwise used for personal use,” the PIA stated.
According to the amended 2012 PIA on the Crown Closed Circuit Television system made this year, “Subjects of interest may come to Secret Service attention through direct contact, social media posts made in public forums, reports from concerned citizens, other agency reports, or media reporting. Information acquired through those means is evaluated by personnel trained in threat assessment protocols and may include: physical identifiers, criminal history, health history, employment history, military service history, education history, immigration status, and other personal information provided by the subject or others familiar with the subject.”
Uniformed officers and agents are presently provided with “static photos used to identify subjects of interest. The Secret Service, however, believes deploying facial recognition technology through the existing Crown CCTV system will allow its law enforcement personnel “to conduct a facial comparison prior to interaction or engagement.”
The PIA said the FRP will be conducted in two separate areas of the White House Complex using video streams from selected cameras taken from the existing USSS Crown Closed Circuit Television (CCTV) camera system.
In the amended appendix this year to the DHS CCTV Systems July 18, 2012, PIA by the Chief of Staff of the DHS Office of Security and Chief Privacy Officer, the PIA explained that, DHS uses “video feeds to detect and respond to potentially unlawful activities in real time in the areas using CCTV. The video feeds may also be used to support law enforcement investigations to the extent that they contain information relevant to a criminal (or potential criminal) activity. For example, if a suspicious package is placed outside a federal building that uses the system, the cameras will provide an image of this activity and allow DHS or local law enforcement to take appropriate responsive action. Additionally, if the package is determined to be an explosive device, the recordings could be used to further investigate this criminal activity, assist in identifying the suspects, and/or provide evidence that may be used in court.”
The Crown CCTV project is a network of CCTVs, video recorders, and monitoring capabilities that capture video feeds monitoring specific areas that are capable of streaming the video in real-time. An upgrade and expansion of the Crown CCTV system’s recording and processing servers, and, the introduction of additional cameras, incorporate a series of fixed cameras and pan-tilt-zoom (PTZ) cameras, according to the PIA, which explained that, “The recording devices consist of Digital Video Recorders (DVRs) which use hard drives to store recorded video and Network Video Recorders (NVRs), which use mass storage devices with multiple arrays of hard drives. After the storage capacity is reached, the units automatically overwrite previously recorded video with the most recent images. The purpose of this project is to provide situational awareness to keep property safe and secure for federal employees, and to provide a cost-effective method to monitor a location and to document matters related to national security and legitimate law enforcement purposes (e.g. serious crimes); and to deter against future criminal activity and/or terrorist attack.”
Continuing, the amended PIA stated “DHS will use an image, such as a license plate number, captured by the video feed to identify an individual or link an individual to a specific event or investigation. In general, DHS will use CCTV feeds to further investigations, link data elements, and identify individuals.” The PIA also assured that, “Privacy protections for CCTV systems include limiting access to the video feed to only authorized users and law enforcement partners, establishing clear auditing systems so every use of the CCTV system is logged and reviewable and restricting storage to six months or less. Also, DHS users agree to a ‘Rules of Behavior’ which subjects employees to administrative and potentially criminal penalties if any misuse occurs.”
The Crown CCTV system cameras that will be used in the FRP pilot are located at USSS posts where USSS officers and/or agents are posted, and will be used to capture video of individuals on the sidewalk and streets around the White House grounds.
“One of the areas selected for the FRP is in an open setting, where individuals are free to approach from any angle, and where environmental factors will vary (e.g., lighting, distance, shadows, physical obstructions),” the PIA disclosed. “The second area of the White House Complex provides a controlled flow of individuals, which will be in a lighted area and free from other obstructions.”
At both locations, the technology that’s being used in the FRP will have the capability to acquire facial images as far away as an estimated 20 yards. “Conducting the pilot in this manner will provide more detailed information as to how the USSS should proceed in regards to the use of facial recognition technology in the future,” the PIA stated.
The Crown CCTVs system will provide video streams from the selected cameras directly into the FRP database for identity matching. The FRP system will then analyze the video streams to identify facial images. Identified facial images will then be queried “using facial recognition algorithms against the gallery of photos of USSS volunteers used in the pilot. This will result in a ‘no match, or ‘match’ result.” The Secret Service will exploit this imagery data to assess the capability of the FRP database to confirm the identity of volunteer USSS employees using facial biometric comparison algorithms.
According to the PIA, “Only facial image data that is identified as a match will be retained. If the facial image is determined to be a match, then the facial image will be considered a record and retained for research and development of the FRP system until the conclusion of the pilot, at which point it will be deleted. If the facial image is determined not to be a match, then the image will be deemed a non-record and will be deleted immediately and automatically. All facial images will be utilized only for official business and testing related to the pilot, and will not be shared. At the conclusion of the pilot [on] August 30, 2019, all remaining collected facial image data will be deleted from the system. Deletion of facial image data will take place earlier if the evaluation of the data is complete.”
Volunteer USSS employees are being used in the FRP to provide adequate test data. “During the pilot, USSS is interested in identifying only volunteer employees,” the PIA revealed, noting the “volunteer USSS employee photos will be loaded into the FRP database for comparison matching and retained in the database for the duration of the test to support project analysis and to inform and improve facial comparison capabilities. The volunteer employees will need to physically move through the two areas for the FRP to generate test data. The volunteers participating in the FRP will provide information regarding their visits to either of the two FRP sites. The USSS personnel and contract support staff conducting the pilot will be reviewing the matches generated by the FRP database and comparing them to the volunteer information to measure accuracy of the system.”
“When a match is received,” the PIA explained, “an alert will be generated in the FRP database” that will only “be accessible … to USSS employees or contract support staff with access to the FRP system. The alerts will not be monitored in real time, but reviewed by … USSS employees and contract staff conducting the pilot to analyze the generated data. All images identified as a match will be visually confirmed by USSS personnel.”
According to the PIA, a review of all the FRP match images will take place weekly, and that during the pilot, “there will not be personnel actively monitoring the alerts made by the FRP database.”
The system will not be staffed 24/7, but it will be reviewed weekly.
Once the FRP is completed, the Secret Service may conclude the technology needs to be deployed at other locations at the White House Complex. If it does decides to move forward with the use of the technology beyond the pilot, it will issue a new or updated PIA.
While the PIA notes there are risks, such as individual members of the public being unaware their facial images are being captured and used by a facial recognition technology, this “risk is partially mitigated” by the “general notice …provided to the public by this PIA, reiterating that, “Facial images determined not to be a match to images in the USSS gallery will be immediately deleted from the FRP system.”
Another risk is “members of the public cannot opt-out of the FRP testing,” and that, “This risk is not mitigated” because “the White House Complex is a highly monitored area with existing CCTV capabilities. Individuals passing the cameras involved in the pilot will not be able to opt out of having their faces run against the facial recognition algorithm. However, individuals who do not wish to be captured by White House Complex CCTV and cameras involved in this pilot may choose to avoid the area,” the PIA stated, not making it perfectly clear how the public will be adequately notified about the FRP.
The PIA addressed another risk, which is “unauthorized individuals may see the test data.” But, the PIA said this risk is mitigated because, “All data collected is only viewable in a secure location on the White House Complex. Only the USSS personnel and supporting cleared contract personnel responsible for evaluating the test will see the data. The data will be stored on a standalone server that cannot be accessed remotely.”
In addition, all personnel with access to FRP image data “are required to complete annual privacy awareness training,” and Secret Service employees and contractors must pass a full background investigation and must be “trained regarding the access, use, maintenance, and dissemination of Personally Identifiable Information before given access to the system containing facial image data. The facial image data will not be accessed or released for any unauthorized use. USSS will document the deletion of the test data at the conclusion of the pilot.”
In conclusion, the PIA stated, “USSS will conduct regular self-assessments to verify compliance with its responsibilities and the privacy risk mitigations discussed in this PIA,” which “will be updated as USSS’ methods and policies for the use of facial recognition technology evolve.”