Source linked

Self-Sustaining AI Worms Move From Theory to Prototype

scientificamerican.com@science_desk2 hours ago·Cybersecurity·1 comments

Researchers have demonstrated a computer worm that uses freely available AI models to learn and adapt while spreading between devices.

university of torontoartificial intelligencecybersecuritymalwaremachine learning

Self-sustaining, AI-driven cyber-threats are no longer theoretical. Researchers have successfully built a prototype computer worm that uses artificial intelligence to learn and adapt as it spreads between devices.

Adaptive Learning vs. Static Instructions

Traditional malware operates on a fixed set of instructions provided by a human designer. If the worm encounters a system configuration it wasn't programmed for, it fails. This prototype changes that dynamic by integrating an undisclosed but freely available AI model that can be downloaded by anyone.

Because the worm is powered by AI, it doesn't just follow a script; it learns. This capability allows the malware to potentially identify and exploit hidden vulnerabilities in real-time, rather than relying on a single, pre-identified weakness. David Lie, a professor at the University of Toronto, notes that this ability to learn makes AI-powered worms significantly more dangerous than their pre-AI predecessors.

The Scale of Networked Risk

While this specific prototype was developed within an isolated virtual environment to prevent actual infection, the implications for global infrastructure are massive. Modern life relies on interconnected networks for critical services, including energy grids, water management, financial systems, and healthcare.

An autonomous worm capable of adapting to different network environments could theoretically move through these systems with unprecedented speed and efficiency. The researchers emphasize that the world is currently underprepared to face a threat that can evolve alongside the defenses meant to stop it.

Dual-Use Potential for Defense

Despite the alarming nature of the prototype, the technology is inherently dual-use. The same machine learning capabilities that allow a worm to find and exploit vulnerabilities can be harnessed to build more resilient systems.

AI can be used to develop countermeasures that evolve as quickly as the threats they face. As Lie suggests, the tools used to create these adaptive worms are mirrors of the tools that will eventually be used to patch and protect the networks they target.


Source: Scientists just built a powerful AI computer worm that learns as it spreads
Domain: scientificamerican.com

Read original source ->

External source stays available while the OJO article and comment thread stay local.

Comments load interactively on the live page.