New AI Worm Poses a Potentially Unstoppable Cyber Threat
▼ Summary
– Researchers have developed an AI-powered worm, named Morris II, that can autonomously spread between AI systems and steal data.
– The worm exploits generative AI tools by using adversarial prompts to infect systems and replicate itself through email clients.
– It poses a “fundamentally new threat” because it can evade current security measures designed for traditional malware.
– The worm was tested on AI assistants like ChatGPT and Gemini, demonstrating its ability to extract personal information and launch spam campaigns.
– Experts warn that this worm could become unstoppable if it spreads widely, as it leverages AI’s learning capabilities to adapt and persist.
Researchers have unveiled a computer worm powered by artificial intelligence that could represent a turning point in cybersecurity. This new breed of malicious software, described by experts as “a fundamentally new threat,” leverages generative AI to spread autonomously and adapt in real time, potentially making it nearly impossible to stop.
Unlike traditional worms that rely on fixed code and predictable behaviors, this AI-driven worm can generate its own variants, rewrite its attack strategies, and evade detection by learning from security defenses it encounters. It uses large language models to craft convincing phishing messages, impersonate trusted contacts, and even manipulate other AI systems into granting access. The worm can also analyze its environment and modify its payload to exploit specific vulnerabilities, a capability that marks a significant escalation in cyberattack sophistication.
Security researchers warn that this development blurs the line between automated malware and targeted, human-level attacks. Once released into a network, the worm could spread through email clients, messaging apps, and other connected services without requiring manual intervention. Its ability to self-improve means that even if one variant is neutralized, another may emerge with updated tactics.
The implications are profound for organizations and individuals alike. Traditional antivirus and signature-based detection methods are largely ineffective against a threat that changes its own code and behavior. Instead, defenders will need to rely on AI-powered security systems that can anticipate and counter the worm’s moves in real time. This creates an escalating arms race where both offense and defense are driven by machine learning.
While the worm has so far been demonstrated in controlled research environments, the underlying technology is already accessible. The researchers who developed it have published their findings to raise awareness and encourage the development of countermeasures. Their work underscores a critical reality: as AI becomes more capable, so too do the tools of cybercriminals. The question is not if such a worm will appear in the wild, but how quickly the security community can adapt.
(Source: Google News)
