ACLU report: US police use AI-flagged travel patterns to stop drivers

The American Civil Liberties Union (ACLU), a US civil rights group, published a report on Monday, October 26, 2026, warning that American police are increasingly using artificial intelligence and data from automatic licence plate readers to flag drivers whose travel histories look "suspicious", even when there is no evidence that they have committed a crime. The report was written by Jay Stanley, a senior policy analyst in the ACLU's Speech, Privacy, and Technology Project. It argues the practice could lead to more "pretextual" stops, in which an officer cites a minor traffic offence while the real reason for the stop is what the surveillance data suggested.

Automatic licence plate readers are cameras, fixed on poles or mounted on police cars, that photograph every passing number plate and store the time and place. On their own they are used to find stolen cars or vehicles linked to a known crime. What has changed, the ACLU says, is the addition of analysis software that looks across millions of these scans and picks out patterns: a car that makes repeated long trips along a known trafficking route, for example, or that visits certain places at unusual hours.

The civil liberties concern is that this turns ordinary travel into grounds for suspicion. A pattern that looks odd to an algorithm may simply be a nurse working night shifts, a family visiting a relative in another state or a delivery driver. The ACLU argues that such systems will sweep in far more innocent people than wrongdoers, and that the person stopped usually has no way of knowing that software, not a human observation, prompted the stop. Because the officer can cite a lane change or a broken tail light, the role of the data may never appear in any record.

Supporters see it differently. Police departments and camera companies say the systems help solve crimes, and that patterns across state lines are exactly what investigators need. Flock says its tools supported more than one million criminal investigations and incidents in 2025, a figure that is the company's own. The debate in the US is about where to draw the line: whether "suspicious pattern" alerts should be allowed at all, and if so with what warrants, audits and limits on keeping data.

The question is not only an American one. Camera-based traffic enforcement and number-plate recognition are spreading in many countries, and how long such data is kept and who may analyse it are questions data protection laws, including India's Digital Personal Data Protection Act, will increasingly have to answer.

Short-term: The report adds pressure on US cities and states that are deciding whether to sign or renew contracts for plate-reading cameras. Local councils and courts may be asked to examine how alerts are used before stops.

Long-term: The outcome could shape rules on AI in policing across the US, such as requirements to disclose when software prompted a stop, limits on how long plate data can be stored, and audits of how accurate pattern alerts are.

Who is affected: Drivers in US communities that use plate readers, police departments, surveillance technology companies, civil liberties groups, and lawmakers writing rules for AI in public safety.

Questions and Answers