Major warning from Google: AI agents are now carrying out cyberattacks without human intervention, increasing the threat to companies..
While Artificial Intelligence (AI) has made tasks easier for people, it is simultaneously becoming a dangerous weapon for cybercriminals. Previously, hackers used AI for minor, routine tasks, but the situation has now shifted. Tech giant Google has issued a warning that cybercriminals are increasingly employing "AI agents"—autonomous systems capable of identifying software vulnerabilities and devising attack strategies without any human intervention.
In its report, "GTIG AI Threat Tracker: Agentic Threat Activity Crosses a Significant Threshold," Google Threat Intelligence Group (GTIG) identified a multi-agent framework dubbed "exploit_agent." This system can largely automate the entire process of discovering and exploiting software vulnerabilities, marking a significant evolution in the methodology of cyberattacks.
How does this AI-powered system operate?
According to Google’s report, "exploit_agent" continuously monitors popular software and its latest updates. When a company releases a security update, the system compares the old code with the new, allowing it to pinpoint exactly which security vulnerability has been patched.
The AI can then devise plans to target organizations that have yet to apply these updates. In essence, the very vulnerability that a company has addressed through an update becomes the key to an attack for hackers.
Google discovered this framework on a server with inadequate security measures. The server was connected to several large-scale and local AI models. It also contained information related to cyberattacks, scripts for network reconnaissance, and data pertaining to other malicious activities. The report notes that this system showed a particular interest in French websites associated with the media, entertainment, and education sectors.
AI Accelerates Cyberattacks
Google has noted that while the methods used by hackers are not entirely new, AI has significantly enhanced their capabilities. According to John Hultquist, Chief Analyst at GTIG, cyber hackers continue to target vulnerabilities in legacy security systems that companies have failed to patch.
The report also highlights two security vulnerabilities in the PaperCut print management server. Researchers, assisted by AI, were able to develop functional exploits for these flaws—which surfaced in late August 2026—within a single day. Shortly thereafter, malicious activity targeting hundreds of organizations was detected.
Google believes that companies must now increasingly leverage AI for cybersecurity. Since attacks are being executed at machine speed, equally rapid technology is required for defense.
Disclaimer: This content has been sourced and edited from Money Control. While we have made modifications for clarity and presentation, the original content belongs to its respective authors and website. We do not claim ownership of the content.

