Home / Technology / AI Worm Learns, Spreads Autonomously
AI Worm Learns, Spreads Autonomously
3 Jun
Summary
- AI worm autonomously exploits unique device vulnerabilities.
- Experimental AI worm is low-cost and uses open-source models.
- AI worms pose new threat to interconnected digital systems.
A new type of AI-powered computer worm has been demonstrated by researchers, capable of autonomously learning and exploiting unique vulnerabilities in devices. This "fundamentally new threat" utilizes an open-source AI model to tailor attack strategies to each specific target it encounters.
Unlike traditional worms, which often exploit a single known flaw, this experimental worm dynamically detects and uses diverse tactics to propagate across networks. It parasitically consumes device computing power, an issue exacerbated by the increasing prevalence of AI-optimized hardware like smartphones and laptops.
The researchers' experimental worm, while currently slower than traditional variants, took approximately five days to infect half of the devices in a controlled network. They warn that this timeframe will likely decrease as devices become more efficient and AI models improve at identifying security weaknesses.
This development arrives amid growing concerns in the cybersecurity sector regarding AI's potential to discover and exploit vulnerabilities at an unprecedented scale. Companies like Anthropic and OpenAI are exploring AI for cybersecurity defense, while the researchers behind the worm published their findings to alert the global community to this emergent threat.
