US and EU lack of protections on use of border biometrics, reports allege
As the European Union continues to make progress on its projects for digital ID, online protection and governance of AI, a new report claims that AI use for border control is not covered by the proposed AI Act. While in the United States, another report unpicks how the Immigration and Customs Enforcement agency built a vast surveillance apparatus using biometrics and billions of data points from digital records on the identity documents of millions of people.
EU’s proposed AI Act will not regulate AI, biometrics border technology
Despite already spending €341 million (US$354 million) since 2007 on research into artificial intelligence technologies for immigration, asylum and border control, the proposed EU AI Act would not regulate such applications of AI and biometrics, claims a report published by Statewatch in a coalition with other human rights organizations.
‘A clear and present danger: Missing safeguards on migration and asylum in the EU’s AI Act’ collates dozens of examples of existing and proposed projects spanning biometric identification and verification devices, automated data-gathering, predictive analysis software, databases and even border control robots. It finds that these are either insufficiently covered or excluded by the proposed AI Act.
The team acknowledges that European Parliament committees have taken a harder line in their assessments of and feedback into the proposed AI Act, but that they have still not addressed the issue of migrants and asylum seekers.
The report also notes that private companies have received the largest portion of funding at €163 million. A previous report by Statewatch found that spending on border enforcement in the EU has increased rapidly since the migrant crisis that began in 2014.
Report builds case against ICE using sophisticated surveillance, biometrics
The U.S. agency Immigration and Customs Enforcement (ICE) “operates as a domestic surveillance agency” accessing information from private companies about the lives of as many people as possible living in America and contracting biometrics firm in its efforts to carry out deportations and playing “a key role in the federal government’s larger push to amass as much information as possible about all of our lives,” according to a new report.
‘American Dragnet: Data-Driven Deportation in the 21st Century’ is the result of a two-year investigation and hundreds of Freedom of Information Act requests as well as a review into how ICE contracts and procures technology, by the Center on Privacy and Technology of Georgetown Law, the law school of Washington D.C.’s Georgetown University.
“In its efforts to arrest and deport, ICE has – without any judicial, legislative or public oversight – reached into datasets containing personal information about the vast majority of people living in the U.S., whose records can end up in the hands of immigration enforcement simply because they apply for driver’s licenses; drive on the roads; or sign up with their local utilities to get access to heat, water and electricity,” states the report.
A review of more than 100,000 spending transactions by ICE finds it spent US$2.8 billion between 2008 and 2021 on surveillance, data collection and data sharing initiatives. This reveals it was building sophisticated surveillance capabilities far earlier than previously thought.
The authors previously believed ICE began conducting facial recognition on state and local data sets from 2013, but have uncovered a 2008 contract with defense and biometrics contractor L-1 Identity Solutions, several years before it became part of what it now Idemia.
This contract allowed ICE to access the facial recognition database of the Rhode Island motor vehicle department in order to “recognize criminal aliens.”
This was the tip of the surveillance iceberg, the report says: “ICE has used face recognition technology to search through the driver’s license photographs of around 1 in 3 of all adults in the U.S. The agency has access to the driver’s license data of 3 in 4 adults and tracks the movements of cars in cities home to nearly 3 in 4 (70 percent) adults.
“When 3 in 4 adults in the U.S. connected the gas, electricity, phone or internet in a new home, ICE was able to automatically learn their new address. Almost all of that has been done warrantlessly and in secret.”
As well as utility and DMV records, ICE also brings in data from social media posts, health care records, child welfare records and geolocation information. It then uses algorithmic tools to search, match and analyze the data.
As recently as August 2020, Clearview AI won a contract with ICE for facial recognition services. Last week ICE extended its contract with Trust Stamp for a facial recognition app for tracking asylum seekers on the U.S.-Mexico border.
Research finds that state authorities are largely unaware of ICE’s surveillance of their residents, while some agencies deny responsibility for ICE access to their records.
The report authors make many recommendations such as that Congress “conduct aggressive oversight of ICE surveillance” to understand how ICE uses biometrics, including facial recognition, fingerprints and DNA and that ICE should end its dragnet surveillance and use of facial recognition on DMV data for immigration enforcement.