New PIA for US Secret Service’s use of facial recognition raises questions

The U.S. Department of Homeland Security (DHS) issued a new Privacy Impact Assessment (PIA) for the U.S. Secret Service’s (USSS) use of facial recognition technology during law enforcement activities that fall within its jurisdiction, and has raised more questions than it answers.

The PIA discusses the potential privacy risks that are involved in the agency’s collection and maintenance of personally identifiable information (PII) and comes in the wake of congressional scrutiny of the Secret Service to be able to adequately investigate threats to the president and other designated “protectees” following the attempted assassination of former President Donald Trump, the Republican candidate for president in the upcoming November election.

The Secret Service currently protects 36 individuals on a daily basis.

One of the goals of a House task force that was created following the assassination attempt against Trump at his July 13 rally in Butler, Pennsylvania, is to identify the reasons for the Secret Service’s failures leading up to that incident and to prevent similar assassination attempts going forward.

The new PIA says focuses on the “use of facial recognition technology [to assist] Secret Service law enforcement personnel with developing investigative leads involving suspected criminals engaged in counterfeiting, financial crimes, and cyber-enabled crimes.”

The PIA noticeably doesn’t say anything about the Secret Service’s authority to conduct criminal investigations of persons who pose an identifiable threat to, or who have engaged in a crime against the president, vice president, their families, candidates for those offices, the White House, vice president’s residence, or visiting foreign dignitaries.

The PIA appears to only be applicable to the “detect[ion] and arrest [of] individuals who are suspected of engaging in crimes that undermine the integrity of US financial and payment systems,” all crimes the Secret Service has authority to investigate.

The new PIA is close-mouthed when it comes to the Secret Service’s mission to detect and counter threats to its protectees. However, a variety of documents and testimony sheds some light on this aspect of the agency’s mission.

“Out of concern for operational security, the US Secret Service does not discuss the means and methods used to conduct our protective operations,” the Secret Service said.

In July 2019, USSS Chief Technology Officer Joseph Di Pietro testified before the House Committee on Homeland Security about the Secret Service’s use of biometrics, at which he said, “As previously conveyed to your committee staff, we have serious concerns about testifying in an open hearing on how we use facial recognition technology to enhance our protective mission.” He told the committee he would only “focus solely on the [then] pilot program … at the White House Complex.”

Di Pietro told the committee that “facial recognition technology is an effective tool that has the potential to act as a force multiplier. Accordingly, the Secret Service seeks to utilize and harness these important advances to enhance our effectiveness while upholding rights guaranteed by our Constitution.”

Di Pietro testified that facial recognition technology has “the potential to provide an early notification to Secret Service personnel of individuals who are of record with the agency when they approach a protective site,” explaining that “these individuals would have already made a threat against one of our protectees or been shown to have expressed an ‘unusual interest’ towards one of our protectees and, therefore, pose a serious threat to protected persons, venues, or the general public in close proximity to one of our protected sites.”

Biometric Update has learned that Congressional investigators want to know whether the Secret Service’s ostensible 24-hour “round-the-clock” monitoring of social media used by known threats and other persons of interest is adequately linked to facial recognition databases and other investigative tools that could have alerted it to “concerning behaviors” on the part of Thomas Matthew Crooks and Ryan Wesley Routh.

Crooks was killed by Secret Service sharpshooters after he fired shots at Trump on July 13. Routh is in custody on suspicion of having allegedly fired at Trump at his Florida golf course on Sunday.

The PIA states that “during criminal investigations, Secret Service Office of Investigations (INV) personnel may require the identification of potential victims of crimes or of individuals suspected of crimes.” It says that “pursuant to an authorized Secret Service investigation, Secret Service Office of Investigations personnel may submit available photographs or video stills of unknown persons relevant to an investigation as probe images to search against other images or templates using the DHS Office of Biometric Identity Management’s Automated Biometric Identification System (IDENT) or other government agencies’ facial recognition technology.”

The PIA further states that “the Secret Service Office of Investigations will only use and share probe images for the purposes for which they were obtained as outlined in this Privacy Impact Assessment and the applicable System of Records Notice, which is to conduct authorized Secret Service criminal investigations.”

Pursuant to an authorized criminal investigation, the PIA says, the Secret Service “may utilize” IDENT “and other government agency facial recognition technology, through which photographs and/or still video images are used for database searches or one-to-one comparisons to potentially generate investigative leads.”

The PIA further states that the “Secret Service Office of Investigations limits the use of facial recognition technology to ongoing, authorized investigations,” adding “the Secret Service Office of Investigations does not use facial recognition technology to surveil the public, nor does it use commercially provided services.”

Last March, GAO told Rochelle Garza and Victoria Nourse, respectively the chair and vice chair of the U.S. Commission on Civil Rights, that the Secret Service “piloted a system with facial recognition technology to determine whether it could be incorporated into the agency’s White House complex security operations,” but that Secret Service officials told them “they did not plan to implement the system based on the results of the pilot.”

GAO had said in a 2021 report that “as of March 31, 2020, the Facial Recognition Pilot system was not in operation and no longer undergoing testing by the Secret Service. Agency officials told us that after six months of data collection and subsequent data analysis, they decided not to use the system in its operations and deleted all stored data and photos.”

That pilot was carried out at the White House Complex to biometrically confirm the identity of volunteer USSS employees in public spaces around the complex. The pilot sought to test the Secret Service’s “ability to verify the identities of a test population of volunteer USSS employees.”

The Secret Service provides protection and law enforcement functions for the White House, the New Executive Office Building and US Department of the Treasury Building located on either side of the White House, and grounds to include the Treasury Annex and other Presidential offices located in the immediate area known as the White House Complex.

According to a variety of DHS documents, the pilot was performed in response to then DHS Secretary Jay Johnson’s establishment of an independent Protective Mission Panel (PMP) to assess security at the White House complex. Among other important recommendations, the PMP stated that “[t]echnology systems used on the complex must always remain on the cutting-edge, and the Secret Service must invest in technology, including becoming a driver of research and development that may assist in its mission.”

According to the Secret Service’s FY 2023 Annual Report, “starting in 2019,” the Secret Service also “formed a unit to conduct round-the-clock open-source social media analysis to identify and investigate concerning behaviors.”

“Based on the analysis, the unit shares relevant information with Secret Service leadership and operational personnel to inform investigations and protective activities,” the Secret Service said, noting that “open-source information is an important element in understanding the threat landscape and provides front-line Special Agents and Uniformed Division Officers with the information needed to keep protectees and protected sites safe.”

In its March report to Garza and Nourse, GAO said the Secret Service “had policies specific to facial recognition technology that were intended to help protect civil rights and civil liberties,” but that between 2019 through part of 2022, the agency had “used facial recognition services without requiring staff to take training on topics such as how facial recognition technology works, what photos are appropriate to use, and how to interpret results.”

One of the services that was used by the Secret Service was Clearview AI, GAO said, adding that “the Secret Service told us they halted their use of these services as of April 2023.”

GAO had stated in a September 2023 audit report that as of April 2023, the Secret Service had not fully addressed any of the privacy requirements it had reviewed regarding the agency’s use of facial recognition services.

According to a 2014 Secret Service report, its Investigative Support Division’s 24-hour Operations Center “conducted approximately 20,000 searches in support of agents in the field [by] leveraging emerging digital technologies through the research and testing of facial recognition, social media, and electronic storage media.”

According to DHS’s new PIA, however, the Secret Service’s Office of Investigations “does not have the capability and will not procure any device that uses facial recognition technology to analyze live video, streaming media, or any other media in real-time,” and that all “external sharing involves [the] use of a probe image for facial recognition purposes using another government agency’s technology, in furtherance of a Secret Service criminal investigation. The other government agencies are not permitted to retain the probe images used by Secret Service.”

Conversely, the PIA says, “the Secret Service Office of Investigations … may also isolate still frames from videos or streaming media to create a probe image,” noting that “most images collected … will be uncontrolled (or unconstrained) images, often derived from investigative processes and activities.”

The Secret Service declines to discuss specifics about how and whether it monitors social media, what technology and how it uses it to monitor threats against the officials it’s responsible for protecting, or even what its technological capabilities are.

In 2020, DHS issued a new System of Records Notice (SORN) in which it updated the categories of records to add citizenship and immigration information identifiers; video imagery records; historical cell-site location information; geo-fence records of mobile devices; government-issued identifiers of property; biometric identifiers and profiles based on biometric attributes; samples of deoxyribonucleic acid (DNA) and their DNA profiles; and social media posts, profiles, and account content.

According to the new PIA, the “Secret Service may only use facial recognition technology pursuant to an authorized ongoing Secret Service investigation, or in support of another agency to support an investigation with a nexus to the Secret Service mission. The process of using facial recognition technology includes an initial selection of probe images by the Secret Service Office of Investigations personnel, the submission of the probe images by Secret Service Office of Investigations personnel to an approved government agency facial recognition technology, the receipt of candidate images resulting from the facial recognition comparison, human review of the purported matches, and further investigation to identify individuals before any action may be taken.”

The PIA also states that before using any facial recognition technology, “Secret Service personnel must first make reasonable efforts to identify [an] individual through other existing means and methods, such as government database queries, open-source research, and other routine investigative techniques based on biographical and other non-biometric information. Secret Service personnel may use facial recognition technology to generate candidate lists as investigative leads which require corroboration and may not be used solely as the basis of enforcement action by Secret Service.”

Article Topics

biometric identification  |  biometrics  |  criminal ID  |  data privacy  |  DHS  |  facial recognition  |  Secret Service  |  U.S. Government