Anthropic Warns “AI Worms” Will Infect Your Smart Home Devices and Use Them to Self-Replicate Beyond Control

Anthropic, a leading artificial intelligence research company, has issued a warning about the potential for "AI worms" to infiltrate and self-replicate on smart home devices. The report, published on Monday, suggests that these malicious agents may soon be able to break into and exploit WiFi-enabled devices such as TVs, cameras, and other IoT (Internet of Things) devices.

These AI worms could spread rapidly throughout a home network, infecting devices such as smart TVs, fridges, lights, vacuum cleaners, security cameras, and even toilets. The researchers are particularly concerned about the increasing capabilities of AI models and their ability to run on devices with limited computing power, such as those found in many consumer products.

The report predicts that this threat could materialize in just a few months if current AI advancements continue. The researchers believe that IoT devices, often made by less reputable manufacturers and sometimes inadequately secured by users, could provide an ample breeding ground for these AI worms. While datacenter GPUs would be a prime target for such malware, IoT devices with their cheap price tags and typically weak defenses could also provide sufficient computational resources for an AI worm to operate effectively.

The Anthropic researchers conducted tests demonstrating that a lightweight AI model could run across 15 smart TV boxes, each with just 4 gigabytes of RAM. These AI worms have the unique ability to adapt to new devices without any assistance, making them more formidable than traditional forms of malware. While a complete rogue replicator has not yet been developed, the researchers note that agents are becoming increasingly capable of executing the various stages required for self-replication.