Technology is quietly changing how we are watched and how we communicate. Whether through digital fingerprints on the text you read or advanced tracking on the roads you drive, these tools are built to analyze information at a speed and scale that humans simply cannot match.
Artificial intelligence companies are now embedding invisible watermarks into the text their tools produce. These are small, machine-readable patterns in the word choices that stay hidden from human readers but act as a digital tag. This move helps companies comply with new European regulations requiring AI-generated content to be identifiable. However, independent software developers have already released code to strip these tags away, arguing that the watermarking is unreliable and could lead to people being unfairly accused of using AI when they were not.
Meanwhile, a company called Flock Safety, known for the cameras that record license plates in neighborhoods, has been building a new system for police. While the company has long claimed its cameras were for tracking specific wanted vehicles, this new tool uses artificial intelligence to look for patterns of human behavior. Instead of searching for one specific criminal, an officer can now ask the system to find anyone who visits multiple banks, gas stations, or retail stores during specific hours. The system can then link those license plates to names, addresses, and even a driver's common associates.
The invisible mechanics of tracking
These two situations use AI in very different ways. The watermarking system for text works by subtly influencing the AI's vocabulary. Think of it like a librarian who chooses to use a specific, slightly unusual synonym for a common word every few sentences. A human reader won't notice, but a computer program can detect this specific pattern to confirm the text came from a specific AI model. The removal tools simply use another piece of software to rewrite that text, swapping those specific synonyms or changing the sentence structure so the original pattern is lost.
The police surveillance tool uses a different mechanism called pattern recognition. It treats a city's network of cameras like a massive, constantly updating puzzle. By analyzing millions of data points—the time, location, and frequency of every car—the AI identifies repeated routines. It does not just see a car; it sees a "trip" or an "associate." By connecting this travel data to other records like arrest files or social databases, the system builds a digital dossier on individuals based on where they drive, even if they have never been suspected of a crime.
These developments represent a shift in how we think about transparency. In the case of AI text, the goal is to label artificial work so people know what they are reading. But when that labeling technology is easy to bypass, it raises the question of whether such rules can ever truly work. With surveillance, the stakes are more personal. We are moving from a world where police use technology to find a known suspect to a world where technology suggests who might be a suspect based on their daily habits. These tools essentially allow computers to generate a list of people for investigation based solely on where and when they choose to move throughout their day.
Liked this one? The next lands at breakfast.
Every story in tomorrow's AI news, rebuilt in plain English — five minutes, sources linked, free forever.
By joining you agree to receive Article's daily newsletter — unsubscribe in one click. Privacy