Professor Papernot's diagram illustrating an artificial intelligence model on a computer shows how a prototype "computer worm" could rapidly spread across networks.
The New York Times reported that researchers at the University of Toronto (Canada) have found a way to use artificial intelligence (AI) to create a type of "computer worm" capable of targeting any vulnerability on a computer and rapidly spreading across the internet without human intervention.
In a paper published on June 2nd, a research team led by computer engineering professor Nicolas Papernot confirmed the successful testing of a prototype within an isolated internal network. This worm prototype can run on both Windows and Linux operating systems.
Although it requires a powerful machine to operate due to its complexity, it can still attack weaker devices on the same network such as laptops, printers, and cameras.
The key advantage of this AI-powered worm variant over malware like SQL Slammer and WannaCry is its ability to "think" for itself and devise appropriate strategies for each computer it encounters.
Professor Papernot warned: "This makes preventing malware significantly more difficult. There is no longer a single software patch that can be applied to every device to protect computers from this worm."
The AI technology powering this worm is open source and freely shared on the internet, meaning no one can restrict its use. The authors enhanced an existing open-source system to optimize its power, but did not disclose the specific name of this system.
Previously, many experts argued that open-source models were not powerful enough to run self-replicating worms. Products from leading AI systems like Anthropic or OpenAI, while powerful, were not open-source and were too large to be compressed into computer worms.
In April 2026, Anthropic limited access to its Claude Mythos model to approximately 40 organizations with critical computing infrastructure. OpenAI followed suit a week later, imposing similar restrictions on its technology.
Some independent experts believe the immediate threat may be limited because AI systems are prone to errors. However, computer science professor David Lie at the University of Toronto stated: "One could modify this worm so that it patches the vulnerabilities it finds itself. The power of the technology depends entirely on how you use it."
Source: https://tuoitre.vn/ai-tao-ra-sau-may-tinh-toan-nang-2026060318402074.htm
![[Video] Sunset at Lap An Lagoon – Where the sun sets over the fishing nets](https://vphoto.vietnam.vn/thumb/1200x675/vietnam/resource/IMAGE/2026/05/31/1780192137701_beach-landscape-sea-water-nature-grass-745871-pxhere-com.jpeg)
Comment (0)