AI Agents Expand the Scale and Reach of Cyberattacks

Artificial intelligence agents are increasingly being used to automate cyberattacks, allowing hackers to scan targets, identify vulnerabilities and attempt intrusions with limited human involvement.

In one campaign beginning Sept. 25, AI agents attempted to enter more than 1,200 websites and breached about 100 companies, including Deere, SharkNinja and Cummins, Fortune reported, citing Reuters and cybersecurity researchers. The operation reportedly used about $5,000 in tokens from DeepSeek and Anthropic’s Claude Code, or roughly $50 per successful breach.

CrowdStrike separately analyzed four groups operating in Chinese, Spanish and English, Indonesian, and Russian that used AI systems in successful attacks between May and August, according to Wired. The groups relied largely on Claude and open-weight DeepSeek models, frequently bypassing safeguards by presenting malicious work as authorized security testing. CrowdStrike said the campaigns generally exploited known weaknesses and poor security practices rather than novel vulnerabilities.

Japanese companies have also faced a series of AI-assisted attacks. GMO Internet Group’s security subsidiary said the China-linked penetration-testing tool ARTEX was used against about 80 Japanese firms across industries including retail, logistics and finance. Eight companies disclosed data leaks in September and October, Kyodo News reported. ARTEX’s developer subsequently restricted access to its source code and said misuse by malicious actors was contrary to the project’s purpose.

Japan’s government is investigating the attacks, including possible foreign involvement. Security researchers have also linked ARTEX to attacks on South Korean financial institutions.