AI is already powering new, more adaptable forms of cyberattack, but hackers typically wield the technology from afar. What if they set the AI models themselves loose to self-replicate and spread in a malicious manner?
Recent experiments have demonstrated two different ways this could happen. Two independent teams showed how a cyberattacker could turn an AI model into a “worm,” a type of malware that, unlike a computer virus, spreads on its own. And AI unlocked a new worm power: adapting attacks to different devices.
The results hint at a new way AI could be used to ramp up cybersecurity threats — not just to computer systems but to smart devices such as security cameras or refrigerators. The good news is that the world still has time to build up defenses as several obstacles remain before this type of malware would work in the real world, says AI researcher Jonas Guan.
To read more, click here.