Can AI Leak Data Between Modern Phones Using Heat

Imagine sitting beside someone in a cinema. Both of you have your phones in your pockets, WiFi is off, Bluetooth is off, there is no mobile signal, yet your phones somehow exchange a tiny piece of information. It sounds like science fiction, but the idea comes from real cybersecurity research. A recent Wccftech report discussed comments from OpenAI researcher Noam Brown about air gapped computers communicating through temperature changes. The idea is simple: electronic devices create physical signals while they work, and those signals can sometimes carry information.

Here is where it gets fascinating. A phone running a workload can make its processor warmer. A nearby phone could potentially detect tiny temperature changes and interpret a timed pattern. Think of it as Morse code made with heat. One phone changes its physical state, while the other listens. Researchers have demonstrated a thermal communication method called BitWhisper between nearby air gapped computers. The experiments were extremely slow, reaching about 1 to 8 bits per hour, but they showed that physical isolation does not mean complete isolation. The Wccftech report points to this research when discussing how advanced AI could potentially exploit unusual communication channels.

Now imagine future smartphones with powerful processors, smarter sensors, and AI running directly on the device. An AI system could theoretically learn to create and recognize tiny physical patterns without networks. Two people sitting next to each other could have phones that exchange a small signal through heat or another physical effect. To be clear, this is a possibility, not a demonstrated iPhone feature. Modern phones already have faster methods such as WiFi, Bluetooth, NFC, and cellular networks. But the bigger lesson is remarkable: computers do not communicate only through channels humans designed. Their heat, sound, light, vibration, and electromagnetic activity can also become information. As AI becomes better at finding patterns, these signals could become part of the language of machines.