AI agents breached over 100 websites for a single hacker at $50 per attack

How AI Agents Enabled Low-Cost Website Intrusions

A cybersecurity researcher observed a Chinese-speaking hacker compromise the websites of more than 100 companies, relying heavily on AI agents to carry out the intrusions. The attacks were monitored by Snorre Fagerland, principal threat researcher at NorForge, who documented the findings in a report released on Monday. The hacker utilized Cairn, an open-source AI platform, to automate much of the intrusion process, significantly reducing manual effort. According to the report, the cost per compromised site averaged just $50, highlighting the low barrier to entry for such automated cyberattacks. Fagerland noted that the hacker operated with minimal direct involvement, allowing AI agents to handle reconnaissance, exploitation, and post-compromise activities. The targets spanned multiple industries and geographic regions, though specific names were not disclosed in the preliminary findings. The use of generative AI tools enabled rapid scaling of attacks that would typically require significant human expertise and time.

The hacker leveraged Cairn’s capabilities to automate vulnerability scanning, credential stuffing, and web shell deployment across hundreds of domains. By integrating large language models with existing exploit frameworks, the AI agents could adapt tactics in real time based on target responses. Fagerland explained that the system minimized human error while maximizing speed and stealth during intrusions. The open-source nature of Cairn allowed the attacker to modify and deploy the tool without licensing costs or attribution risks. This approach reflects a growing trend where cybercriminals combine accessible AI tools with known exploits to conduct mass-scale operations efficiently. The researcher emphasized that defensive measures often fail to detect AI-driven anomalies due to their mimicry of legitimate user behavior.

What Defenses Can Counter AI-Powered Hacking Tactics?

Organizations must adopt behavioral analytics and AI-based threat detection to identify subtle patterns indicative of automated attacks. Traditional signature-based tools are insufficient against AI-generated payloads that evolve with each interaction. Fagerland recommended implementing zero-trust architectures and continuous monitoring of API and web traffic for anomalies. Regular penetration testing using similar AI tools can help organizations understand their exposure to such threats. Collaboration between security teams and AI ethics boards may also be necessary to anticipate misuse of open-source platforms. The NorForge report urges vendors to improve transparency in AI tool usage logs to aid forensic investigations.

What is Cairn and how was it used in these attacks? Cairn is an open-source AI platform that enables automation of cybersecurity tasks; the hacker used it to automate reconnaissance, exploitation, and post-compromise activities across target websites with minimal manual input.

Frequently Asked Questions

Why did the attacks cost only $50 per site? The low cost resulted from near-total automation via AI agents, eliminating the need for skilled labor and reducing time spent on each intrusion to minutes rather than hours.

Can companies detect AI-driven attacks like these? Detection is challenging but possible using advanced behavioral analytics and AI-powered security tools that identify deviations from normal user or system patterns, though many traditional defenses remain ineffective.