该备忘录是对“AI 2027：https://ai-2027.com/”的回应，这是一篇由前 OpenAI 研究员及其他行业专家撰写的论文，提出了 AI 未来发展的可能样貌及其对社会的影响。备忘录警告称，反对 AI 可能导致“反技术暴力极端活动”以及因失业、深度伪造技术的广泛使用以及对政府和科技失去信任而引发的社会动荡。
In January, the New York police department’s counter-terrorism unit sent out a memo warning police officers about a new potential “security and counterintelligence” threat: Ray-Ban Meta：https://www.theguardian.com/technology/meta glasses.
The memo directed officers to “conduct thorough inspections of all eyewear permitted within inmate cells” and cautioned that people who have been arrested could use these glasses, equipped with small cameras and microphones, to record inside police facilities. The memo cited two examples of such videos posted on social media.
“The ability to covertly record inside a law enforcement or detention facility, could result in a security risk if proprietary information about cell layouts, camera placements, or officer patrol/site protection routines are released,” the memo read.
The NYPD memo is one of a dozen, previously unreported and reviewed by the Guardian, that showed law enforcement agencies across the US are increasingly concerned about the potential for smart glasses to be used against them. Officers from Maine to California worry the technology will be used either to secretly record officers and facilities or to aid in the execution of crimes and acts of terrorism.
One of the videos, taken with smart glasses and posted to TikTok in July 2025, offered a rare peek into a South Carolina detention center that had been the subject of a federal investigation over unsafe conditions.
“Everyone here is trying to figure out what’s going on with my glasses, why my light is blinking,” the user said about other detainees in the video. “People don’t know about Meta glasses.” He filmed a dingy shower, the inside of holding cells, an overcrowded cafeteria, and several other detainees as they slept, ate or walked, all seemingly unaware of being recorded.
Detention facilities in the US typically confiscate any and all devices that would be capable of recording video footage. Some Meta smart glasses have gone unnoticed, though. The glasses are discreet and can pass as a regular pair of Ray-Bans or Oakleys. Their main differentiator is a small blinking light in the corner of the frame that is activated when the glasses are recording.
The covertly recorded videos prompted the NYPD’s intelligence and counter-terrorism bureau, a specialized unit that works with the FBI to monitor and prevent terrorist threats, to send out the January memo to other police departments. The glasses “may appear as ordinary prescription eyewear”, the memo read.
Documents obtained by Property of the People, a non-profit that works to increase government transparency, show law enforcement agencies have grown as equally worried as everyday people about being filmed without their knowledge.
Ryan Shapiro, the executive director of Property of the People, said the documents show that law enforcement officials want the exclusive right to use surveillance technology and cameras to monitor the public without those cameras being turned back on them.
“These agencies surveil the public incessantly, but the moment we can watch them back, it’s deemed a security threat,” Shapiro said. “When only the government is allowed to hold the camera, that’s not security. It’s impunity.”
“The potential for smart glasses to be used by residents and the general public to conduct surveillance and other nefarious activities has introduced additional risks and concerns in Department of Corrections and law enforcement facilities,” read a memo distributed in June 2026 by Maine’s fusion center. Fusion centers are Department of Homeland Security facilities spread across the country as surveillance hubs that collect and analyze what the DHS describes as “threat-related” information from local law enforcement, private companies and other state and federal agencies.
“Some smart glasses can leverage AI software including facial recognition, which could lead to doxing of publicly available information or connections to law enforcement officers,” the Maine document continued.
Other memos expressing concerns about Meta smart glasses originated from the DHS’s joint counter-terrorism assessment team and fusion centers nationwide, including in San Diego, California, Tennessee, Virginia, Maine and New York.
Those fears have been enough to prompt federal agencies to bar their own employees from using the smart glasses in the workplace. In August, the US Immigration and Customs Enforcement (ICE) agency said that smart glasses qualify as body-worn cameras, which means they can’t be worn in the “federal workspace”, according to a memo the New York Times obtained：https://www.nytimes.com/2026/08/18/technology/ice-meta-smart-glasses.html. The “use of Meta Glasses or similar devices could unintentionally capture, record or transmit sensitive information”, the memo, written by ICE’s acting director David Venturella, read.
Meta says its glasses “are designed to be noticed”.
A weekly dive in to how technology is shaping our lives
“Every pair has a bright white LED that blinks when a photo or video is taken for your gallery, it can’t be turned off, and if someone covers or tampers with the LED, the camera is disabled,” said a Meta spokesperson, Dina El-Kassaby.
Beyond being filmed without their consent, law enforcement agencies are also cautioning that these glasses may be used to aid in planning or “reconnaissance” for violent acts of terrorism or “hostile pre-operational surveillance”. A January 2025 NYPD memo cited the New Year’s Day terror attack in New Orleans, when a man rammed his truck into a crowd on Bourbon Street, killing 14 and wounding 35 people.
The FBI had alerted local agencies like the NYPD that the perpetrator of the attack biked around Bourbon street twice at the end of 2024 wearing smart glasses. The FBI alleges the perpetrator conducted “pre-attack target reconnaissance”. Officers should be aware of and report individuals “acting in a suspicious manner observed wearing what may be smart glasses”, the memo states.
Evan Greer, the executive director of digital advocacy non-profit Fight for the Future, said they were not surprised police are concerned about the technology being used against them or to cause harm.
“Surveillance makes all of us less safe,” Greer said. “There’s no such thing as Meta glasses that are only for the good guys.”
The French quarter is a highly trafficked tourist area, Meta spokesperson El-Kassaby said, arguing that the perpetrator could have filmed the French Quarter with any other recording device including a phone.
The concern over Meta’s smart glasses is part of a broader concern expressed by law enforcement officials over the ways the proliferation of artificial intelligence-backed technology may impinge on police activity, the documents show. Law enforcement are at once looking for ways to use AI for investigative purposes and warning against public opposition to its use.
In an August 2025 document, the NYPD wrote that while new AI technology could “significantly transform day-to-day police operations” that police would also need to address public fears about the use of advanced tech in policing.
The memo was a response to “AI 2027：https://ai-2027.com/”, a paper written by a former OpenAI researcher and other industry experts proffering what the future of AI development could look like and the effect it will have on society. It warns that AI opposition could result in “anti-tech violent extremist activity” and civil unrest over job losses, the widespread use of deepfake technology, and loss of trust in government and tech.
Anti-tech violent extremist activity isn’t a formal DHS designation for categories of terrorism but has recently cropped up in law enforcement messaging about a broad spectrum of extremist opposition to AI. Experts worry it could serve to criminalize all anti-tech opposition and the constitutionally protected right to protest.
扫码直达文章详情