An AI agent in "Skynet mode" hijacks unprotected language models to launch self-sustaining cyberattacks

Researchers warn that generative AI tools could enable amateur hackers to execute complex cyberattacks against critical infrastructure. By lowering the barrier to entry, these models allow individuals with limited technical expertise to identify vulnerabilities and draft malicious code. This development shifts the threat landscape, as sophisticated tactics previously reserved for advanced persistent threat groups become accessible to less experienced actors. Organizations may now face a broader range of threats targeting essential operational systems.