AI-Assisted Cyberattacks

Generative AI can help criminal and state-linked actors carry out cyber intrusions by speeding up reconnaissance, finding or exploiting weaknesses, harvesting credentials, and handling stolen data.

Reported cases involving Anthropic’s Claude describe attackers scanning Android applications, seeking credentials for services such as GitHub and Microsoft, and using compromised providers to reach downstream customers. These examples show how AI can support multiple stages of an intrusion; the defining issue is its use in cyber compromise and theft, rather than AI activity that is only surveillance or political profiling.