AI is unmasking ICE officers. Can Washington do anything about it?

A new twist in the debate over surveillance tech raises tough questions for policymakers.

An activist has started using artificial intelligence to identify Immigration and Customs Enforcement agents beneath their masks — a use of the technology sparking new political concerns over AI-powered surveillance.

Dominick Skinner, a Netherlands-based immigration activist, estimates he and a group of volunteers have publicly identified at least 20 ICE officials recorded wearing masks during arrests. He told POLITICO his experts are “able to reveal a face using AI, if they have 35 percent or more of the face visible.”

The AI-powered project adds a new twist to the debates over both ICE masking and government surveillance tools, as immigration enforcement becomes more widespread and aggressive.

ICE says its agents need to wear masks to prevent being unfairly harassed for doing their jobs. To their critics, agents in masks have become a potent symbol of unaccountable government force. The masking, and the counter-campaign to identify agents, has prompted a crossfire of bills on Capitol Hill.

ICE agents “don't deserve to be hunted online by activists using AI,” said Sen. James Lankford (R-Okla.), who chairs the Senate Homeland Security subcommittee on border management and the federal workforce.

Some Democrats concerned about the masking are pushing for regulations to make it easier to identify law enforcement officials — but they still say they’re uneasy that vigilante campaigns have begun using technology to do it.

Sen. Gary Peters (D-Mich.), who co-sponsored a bill called the VISIBLE Act to require ICE officials to clearly identify themselves, has “serious concerns about the reliability, safety and privacy implications of facial recognition tools, whether used by law enforcement … or used by outside groups to identify agents,” an aide told POLITICO.

Skinner’s AI-powered unmasking project is part of a broader online campaign, the ICE List, that has published the names of more than 100 ICE employees, from field agents to back-office bureaucrats. It’s one of several anti-ICE campaigns that have drawn attention from the media and Homeland Security officials.

ICE did not comment on the accuracy of Skinner’s purported identifications, but in a statement, ICE spokesperson Tanya Roman said that the masks “are for safety, not secrecy” and that these listings threaten officers’ lives.

“These misinformed activists and others like them are the very reason the brave men and women of ICE choose to wear masks in the first place, and why they, and their families, are increasingly being targeted and assaulted,” Roman said.

The Department of Homeland Security criticized his ICE List project in a July statement, saying Skinner’s efforts appear to be responsible for doxing federal officers.

In response to efforts to identify ICE agents, Sen. Marsha Blackburn (R-Tenn.), who chairs the Senate Judiciary subcommittee on privacy and technology, introduced the Protecting Law Enforcement from Doxxing Act in June, which would make it illegal to publish a federal officer’s name with the intent to obstruct a criminal investigation.

Blackburn told POLITICO via email that Skinner’s project reinforces the need for her bill: "Those who oppose the rule of law are weaponizing generative AI against ICE agents," she said in a statement, warning it could expose agents to threats from transnational criminal gangs like MS-13.

Her record illustrates the political complexities of modern surveillance. In the past, Blackburn has raised concerns about government use of facial recognition, questioning the IRS’s use of the technology in 2022 and criticizing the Chinese government’s surveillance capabilities.

A Blackburn spokesperson said she is against the public’s use of computer-assisted facial recognition to identify ICE officials but supports police use of the technology.

Under existing U.S. law, however, Skinner’s project is legal — highlighting to lawmakers and law enforcement officials the downside of years of congressional inaction on surveillance and privacy laws.

In the absence of federal regulation, the International Biometrics + Identity Association, a trade group that represents the identification technology industry, published ethical standards for facial recognition providers in 2019, which includes ensuring that people’s biometric data isn’t collected without people’s knowledge and consent.

Skinner said the organization’s guidelines aren’t applicable to his efforts, noting that the ICE List uses facial recognition tools but is not a provider of the technology itself.

He declined to describe what AI model the tool is built on but said the tool generates its best guess on what the officer looks like unmasked, using screenshots from ICE arrest and raid videos.

Skinner sends batches of these artificially created images for volunteers to use on reverse image search engines like PimEyes. The company, which offers facial recognition capabilities to the public, trawls through millions of images posted online, often turning up social media profiles on LinkedIn and Instagram.

PimEyes did not respond to requests for comment.

While the technology is recent, the method is not — and in fact a version of it has been used by American police departments on civilians. A 2019 study from the Georgetown Law Center on Privacy and Technology found police departments digitally altering pictures and using artist sketches as the basis for finding suspects through facial recognition.

Privacy experts had qualms about Skinner’s new twist on this idea: “Regardless of how you use it, it’s a rather unreliable application of the technology when you stop actually scanning the face and start scanning an artificial image,” the Center for Democracy and Technology’s Security and Surveillance Project Deputy Director Jake Laperruque told POLITICO.

Skinner acknowledged that the technology is flawed, and he said that about that 60 percent of the AI-generated results and facial recognition searches lead to wrong matches on social media profiles. He says a group of volunteers verifies them through another process before posting any names online.

Asked about the risk to ICE officers, Skinner said he doesn’t believe that the ICE List is endangering officers, saying he posts only names and does not include officers’ addresses or contact information in the listings. He does acknowledge that a name alone would be enough to lead to an officer’s personal data, but discouraged doxing as undermining his anti-ICE effort.

“I don’t believe in public justice, but I do believe in public shaming and public accountability,” Skinner said.

Although Democrats are proposing legislation that would ban officers from wearing masks and Republicans introduced bills that would make it illegal to dox police, no legislation that would limit public use of facial recognition, or prevent companies from selling officers’ personal information have gained momentum.

Commercially available information makes it simple to buy a majority of Americans’ personal information through just a name, which has put lawmakers, judges and police officers at risk.

Privacy experts suggest that stronger data protections would be more effective for protecting officers from doxing than wearing masks or outlawing posting officers’ names.

“If someone doesn’t want [their information] online, they should be able to get it scrubbed reasonably. That’s what needs to be tackled here, not the idea that law enforcement officers in the performance of their duties can be identified,” Laperruque said.

Sign up for POLITICO Breaking News Alerts to receive the latest updates in your inbox.

(http://www.autoadmit.com/thread.php?thread_id=5767969&forum_id=2).#49225247)

fucking clown

(http://www.autoadmit.com/thread.php?thread_id=5767969&forum_id=2).#49225272)

That about sums it up

(http://www.autoadmit.com/thread.php?thread_id=5767969&forum_id=2).#49225336)