Flock Safety’s OS Investigate platform can identify vehicles, map their movements and rank associated drivers without requiring a name or license plate, prompting new concerns from privacy advocates and lawmakers about warrantless surveillance and AI-assisted policing.
By yourNEWS Media Newsroom
Police departments using Flock Safety technology can now search for and track vehicles based on where they travel and how they move, even when officers do not begin with a driver’s name or license plate number.
According to a Daily Caller News Foundation report, Flock’s OS Investigate system, formerly called Nightshift, uses artificial intelligence to identify vehicles and rank a driver’s potential “associates” based on how frequently vehicles appear together at the same Flock cameras.
The system has drawn growing scrutiny as details emerge about how much information it can combine.
Previous reporting on Flock’s camera network and law-enforcement use of the technology has raised questions about how automated license-plate readers are evolving into broader surveillance systems.
WIRED reported that code on Flock’s website identified 45 functions available through OS Investigate.
Those capabilities included license-plate scans, camera metadata, arrest records, case files and 911 dispatch information that could be used together to identify potential suspects.
Privacy advocates say the concern is no longer limited to police investigating a person they already suspect.
“[T]his [Flock tool] really crosses a line from a tool that’s used to investigate suspicious people to deciding who is suspicious,” Jay Stanley, senior policy analyst with the ACLU’s Speech, Privacy and Technology Project, said.
“[I]t increases the chances that you are going to get put under the microscope by law enforcement because some algorithm thinks your driving patterns are ‘suspicious.’”
Stanley predicted that privacy concerns will increasingly push organizations toward AI systems that operate locally rather than sending sensitive data to outside providers.
“[A] lot of AI use is going to have to be local models,” he told the DCNF.
He said homes and local institutions, including hospitals and police departments, could eventually operate their own AI systems for less demanding tasks as the technology improves.
“I think that’s the direction that a lot of organizations are going to go … I’ve talked to police officers who are like, ‘yeah, I want to use AI, but I can’t use AI in our internal stuff. We can’t let that leave the premises,’” Stanley said.
Flock Safety, the Texas Municipal Police Association, the International Association of Chiefs of Police, Code Four Labs, Tyler Technologies and 3SI Security Systems did not immediately respond to requests for comment from the DCNF.
Other critics are focused less on the existence of AI tools than on how readily officers may rely on their conclusions.
Alex Marthews, national chair of Restore The Fourth, said police officers have generally not received enough training to critically evaluate the output of large language models before acting on their findings.
“[L]aw enforcement have not right now been receiving in-depth training on how to challenge and question the findings of an LLM as to who the top 10 bad guys are in their jurisdiction, and they’re probably not going to wait around a long time to parse it all out carefully,” Marthews told the DCNF.
He argued that such tools could allow law enforcement agencies to outsource parts of their decision-making to AI.
If a department can ask a system to “surface for me, ChatGPT, the top 10 real goddamn asshole problems that there are in my jurisdiction,” Marthews warned, an AI hallucination could potentially place innocent people under suspicion.
Marthews pointed to a previous controversy involving Calvary Chapel San Jose.
The church continued holding indoor services during COVID-era restrictions, and authorities later used phone-location information to monitor people who had been present.
He argued that worshipers could not have anticipated that such technology might eventually be used against them.
The same concern, he said, applies to AI-powered surveillance systems.
“This flock system can be targeted against anyone. Governments change, ideologies change, state enemies change, and we do not know in setting up a system that 10 years down the line we won’t be the targets of it,” Marthews said.
Flock CEO Garrett Langley has defended the company’s technology while arguing that Americans must balance public safety and personal privacy.
“[W]hen people talk about just one of these, privacy or safety, they’re prioritizing the wrong thing. What we have to prioritize as a country is ‘compromise,’” Langley told Fox News in an interview later reposted by Sen. Rand Paul.
“How do we have our safety, and how do we balance privacy.”
Langley later responded directly to criticism from Paul.
“[T]hat was not my comment. We believe in the constitution at Flock,” Langley said in another post. “I hope you can agree that Americans deserve to be safe.”
Flock describes its FlockOS platform, which includes OS Investigate, as a command center capable of operating from virtually anywhere.
The company has highlighted cases in which police used its technology to follow vehicles across multiple locations.
“We watched the vehicles enter through LPR, tracked them across the city, and our teams were in place before the suspects even got out of their cars,” Maj. Patrick Krieg of the Dunwoody, Georgia, Police Department said, according to Flock’s website.
“It was the first time we said: this platform is powerful.”
Some technology developers argue that AI can be valuable in police investigations when it is confined to evidence already collected in a legitimate case.
George Cheng, CEO and co-founder of Code Four, told the DCNF that investigators can be overwhelmed by the amount of material generated by body cameras, jail calls, phone extractions and telephone records.
Code Four’s INSIGHTS system combines evidence from an open case and allows detectives to query that information in ordinary language, with responses tied back to underlying source material.
Cheng said there should be a clear distinction between using AI to analyze evidence in an existing investigation and using it to search broadly through the public in search of potential suspects.
He argued that investigations should begin with the former.
Cheng also said AI-generated findings should cite their sources, officers should remain responsible for whatever software produces, agencies should maintain audit trails showing which work was performed by machines and which was performed by humans, and law-enforcement data should never be used to train outside AI models or sold.
James Czerniawski, head of emerging technology policy at the Consumer Choice Center, similarly argued that the central issue is how governments use AI rather than the technology itself.
“AI is simply a tool. It’s going to get utilized in many ways that are great, and it’ll get utilized in many ways that are bad,” Czerniawski told the DCNF.
“[B]ut we don’t regulate the tool itself, we regulate based on the outcome. In this case, the outcome is that you have people’s Fourth Amendment rights getting violated by the government, and there’s no good path to fixing that right now.”
Czerniawski said warrants provide one of the clearest protections.
“‘Warrants’ is your best friend … for both law enforcement and for people’s due process consideration,” he said.
A warrant, he argued, forces police to justify an AI-assisted search before using surveillance information against an individual.
Political pressure for new restrictions is also growing.
“I think there needs to be legislation, both at the federal and state level,” Knox County, Tennessee, Mayor Glenn Jacobs said in an August interview with the DCNF.
“I think that as this issue becomes more prevalent around the country, Congress will take it up.”
Rep. Keith Self, R-Texas, is attempting to establish a federal warrant requirement through legislation.
“The PRIVACY Act prohibits federal agencies from using output from applications such as Flock Safety’s OS Investigate without first obtaining a warrant,” Self told the DCNF.
“Flock Safety’s camera network is one of several emerging technologies that surveil everyday Americans, which requires Congress to implement processes to anticipate new technologies that must require warrants.”
Self’s proposed PRIVACY Act would also limit retention of information gathered by the cameras to 30 days when a warrant has been obtained, unless the information is evidence in a prosecution or a court authorizes an extension.
“AI has supercharged the surveillance state, trampling our God-given right to be secure in our persons, houses, papers, and effects,” Self said.
“These cameras are no longer just license-plate readers looking for vehicles of interest; OS Investigate shows they are being used to aggregate an expanding amount of personal data, raising constitutional questions and putting individual privacy at risk.”
Self called for Congress to impose new restrictions on technologies used to monitor members of the public where Fourth Amendment protections may be implicated.
Sen. Josh Hawley, R-Mo., has also opened a congressional inquiry.
Hawley launched a Senate subcommittee investigation into Flock cameras on Wednesday, according to a letter he sent to Langley.
Concerns about misuse have also been fueled by documented incidents involving automated license-plate reader systems.
The Institute for Justice identified more than 100 cases of alleged ALPR abuse in a database published Aug. 12.
Among those cases was a Georgia police chief who conducted approximately 600 searches to track an ex-girlfriend and her daughter, according to The Washington Post.
The broader debate now centers on how far law enforcement should be allowed to go when combining AI with vast networks of cameras and government records.
Supporters emphasize the ability to find suspects faster and manage enormous amounts of evidence, while critics warn that systems capable of identifying people through patterns of movement and association could place individuals under investigation without the traditional starting points of a name, plate number or established suspicion.
As Flock’s technology becomes more sophisticated, lawmakers and privacy advocates are increasingly focusing on whether warrants, data-retention limits and other safeguards should be required before AI-powered surveillance tools can be used against members of the public.
Original article
Comments