What happened
A recent WIRED investigation has successfully reconstructed and analyzed the source code for the new AI-driven operating system developed by Flock Safety, a prominent surveillance technology provider for law enforcement. While Flock is primarily known for its Automatic License Plate Recognition (ALPR) cameras, the investigation confirms that their next-generation technology—already deployed by several police departments—possesses significantly broader capabilities. The system can now identify specific vehicle features (such as roof racks or bumper stickers), as well as the presence of pedestrians and animals, effectively turning static cameras into sophisticated behavioral monitoring tools.
Technology context
Flock's technology is rooted in Computer Vision, a branch of artificial intelligence that trains computers to interpret and understand the visual world. Utilizing Deep Learning models, the system does not merely "see" an image; it deconstructs it into searchable metadata.
Instead of searching solely for a string of characters (a license plate), the AI employs convolutional neural networks to classify objects within the frame. For instance, the system can distinguish between a bicycle, a pedestrian, or a dog, and allows police to filter searches by attributes that were previously unidentifiable through traditional methods. This processing often occurs via edge computing, meaning the data is analyzed directly on the device to enable near-instantaneous alerts to police dispatchers.
Why it matters
The implications of this discovery are profound for the balance between public safety and civil liberties.
1. Surveillance Expansion: Moving from vehicle tracking to monitoring people and specific personal attributes creates a pervasive surveillance net.
2. Lack of Transparency: The uncovered code reveals functionalities that the public was not fully informed about during public procurement processes.
3. Efficiency vs. Abuse: While these tools assist in solving crimes rapidly, the ability to search for "a person with a red backpack" across an entire city raises significant risks regarding profiling and misidentification.
Key terms explained
- Computer Vision: A field of AI that enables computers to derive meaningful information from digital images or videos.
- Deep Learning: A subset of machine learning based on artificial neural networks that can learn patterns from unstructured data.
- Metadata: Data that provides information about other data; in this context, it refers to the attributes of an object (color, type, brand) extracted from a video feed.
- Predictive Policing: The usage of mathematical and analytical techniques in law enforcement to identify potential criminal activity.
Impact
In the short term, law enforcement will possess much more powerful tools for locating suspects, potentially leading to higher case clearance rates. However, in the medium term, this will intensify the legislative debate surrounding AI regulation in public spaces. There is a tangible risk that these systems could be used to monitor protesters or marginalized groups, leading to an erosion of public trust in state institutions and AI technology as a whole.
What's next
We can expect Flock and its competitors to integrate Large Language Models (LLMs) to allow officers to query the system using natural language (e.g., "Show me all men who passed here in the last hour wearing blue hats"). The trend is moving toward "predictive surveillance," where AI doesn't just report what happened but attempts to identify suspicious patterns before a crime is committed, necessitating much stricter ethical and legal frameworks.
Sources
- WIRED: "Flock Has a Powerful New AI Tool for Police. We Got Its Code"
- VentureBeat AI / Surveillance Tech Industry Reports
*
Educational analysis generated with AI and editorially reviewed.