Advocates of predictive policing like to emphasize how it uses crime data to anticipate future criminal activity. However, a Sept. 8 report by 404 Media raises questions as to what information the U.S. Border Patrol’s secretive predictive policing program is using to target drivers not accused of any crime. The report, based in part on a DHS memo produced in a criminal case, claims the Border Patrol’s predictive policing unit is analyzing the financial data of Americans to provide intelligence to local police officers who then use purported traffic violations as a basis to stop those drivers.
404 Media is not the first to find that the Border Patrol is using information it has secretly gathered to target drivers based on its predictive policing program. In November 2025, the Associated Press found the Border Patrol’s predictive intelligence system monitors millions of American drivers far from the border using a network of license plate readers to target individuals for suspicious travel patterns. According to the Associated Press, some of the targeted individuals were stopped, searched, and sometimes arrested without being given any indication of why they drew the interest of law enforcement.
A U.S. Department of Homeland Security memorandum of investigation related to the recent arrest of Kyle William Olson said it was based on information generated by the department’s Predictive Intelligence Targeting Team (PITT). 404 Media obtained the memo. The memorandum listed law enforcement records cited by a Border Patrol agent in the Spokane Sector for targeting Olson as including “Information contained within law enforcement-sensitive systems suggesting financial activity patterns commonly associated with illicit narcotics activity.”
The memorandum does not explain how PITT is collecting the financial data or specify what data was collected. 404 Media said Customs and Border Protection did not answer questions related to the monitoring of financial activities by Border Patrol officers or whether it was used to obtain a warrant for Olson.
Predictions Based on Huge Amounts of Data
Predictive policing programs generally refer to the use of predictive analytics, artificial intelligence, and other techniques to identify potential criminal activity by analyzing huge amounts of crime and other data. Unfortunately, due to the secretive nature of the programs, departments and agencies that have implemented predictive policing programs have not often shared the types of information used to make those determinations.
The use of predictive policing in the United States has been controversial. Supporters of predictive policing contend it uses historical crime data to help predict crimes accurately and objectively. Critics of its use note that there is little transparency in the systems and that predictive policing could be used to confirm pre-existing racial biases. Despite those concerns, predictive policing has been implemented in more than a half-dozen states, including Arizona, Alabama, California, New York, Illinois, South Carolina, Tennessee, and Washington.
It is not clear when U.S. police departments began using predictive intelligence, but there is little discussion of the practice being implemented in the United States before 2010. An article in the FBI Law Enforcement Bulletin noted that a study released in 2010 found it to be feasible. The article goes on to explain that by 2013 police in Santa Cruz, Calif., had already implemented a predictive policing program and were using an algorithm supplied by a local college professor to analyze crime data to better allocate police resources.
Also in 2013, the RAND Corporation produced a guide to help police departments with developing strategies to more effectively conduct investigations and prevent crime that discusses predictive policing. The guide observes that predictive policing can’t actually predict crimes. It can only tell law enforcement which people have a higher risk of committing crimes and where those crimes might be committed.
The Arrest of Kyle William Olson
According to 404 Media, Olson was driving on Interstate 90 in Montana at 9:30 with his dog Ducky. He works on a licensed cannabis farm in California and had marijuana edibles, marijuana cuttings, and butane hash oil in his vehicle. Olson claimed to be transporting the drugs back to his home in Wisconsin.
A police report said two Montana Highway Patrol officers were sitting alongside the road when Olson passed. Noticing that Olson’s license plate was obstructed by a tinted cover and a luggage holder, the officers pulled Olson over. According to bodycam footage, one officer told Olson he was stopped because his license plate was “a little bit obstructed.”
The officer demanded that Olson step out of the vehicle and charged him with a DUI after administering a drug test. Olson was later charged with possession with intent to distribute due to the marijuana in his vehicle.
While preparing his defense, Olson was provided with the DHS memorandum (later obtained by 404 Media) that described how the Border Patrol was responsible for instigating the traffic stop. While the document claims the Border Patrol intelligence was not the sole basis for the stop, the arresting officer said that the Border Patrol had identified Olson’s vehicle as having possible indicators of narcotics trafficking.
The Civil Liberties Concerns
Document discovery in the narcotics trafficking case against Olson has revealed that PITT has been using the financial data of Americans, along with information related to past criminal activity, to target drivers when there may not be other indicators of wrongdoing. The information reported by 404 Media, coupled with the Associated Press’s recent finding that the Border Patrol was also using location data from license plate readers far from the border in its analysis, raises questions regarding what other information collected from the innocent activities of Americans is being used as part of PITT’s mass surveillance program.
