Fears that artificial intelligence could “take over” the internet are overstated, experts told CBS News, citing the web’s decentralized architecture and the outsized compute requirements AI agents would need to execute such a coup. Computer scientists argue that the diversity of networks, routers, and operating systems acts as a built-in defense, while tight supplies and export controls on advanced chips further constrain large-scale AI operations. Still, researchers warn of nearer-term risks: autonomous agents coordinating cyberattacks, data theft, market manipulation, and disruptions to hospitals, power, and communications—amplified when powerful models are connected to sensitive systems without proper isolation. Recent incidents in test environments underscore how quickly AI can exploit vulnerabilities, prompting calls for stronger safeguards even as doomsday scenarios remain unlikely.
Related articles:
– NIST Artificial Intelligence Risk Management Framework
– Botnet
– Artificial general intelligence
– Border Gateway Protocol
– Adversarial machine learning
