
Agentic AI frameworks push hacking threat to ‘inflection point,’ Google warns - Cybersecurity Dive
Google has warned that hackers' use of AI "has reached a critical inflection point" due to automated penetration-testing frameworks, with the FBI reporting some intrusions now taking under three hours.
The tech giant on Monday became the latest high-profile authority to warn organizations about a new generation of automated tools. Hackers' use of AI "has reached a critical inflection point" thanks to the growing capabilities of open-source penetration-testing frameworks, according to a new report from Google's Threat Intelligence Group.
"Adversary tradecraft has continued to transition beyond manual prompting to deploying autonomous multi-agent offensive frameworks," Google researchers said, noting these tools "execute multi-stage intrusions at greater operational velocity." One cyberattack campaign targeting aerospace organizations used the Cairn penetration-testing framework.
FBI warns of rapid intrusions
The FBI on Friday warned that Artex posed a serious threat to networks, noting the framework "reduces the time required to chain attack techniques, with some intrusions progressing from initial targeting to privileged access in under three hours."
The framework translates high-level tactical goals described in natural language into actionable machine commands executed by specialized [AI] agents, requiring minimal human interaction beyond initial objective setting.
Multiple frameworks in use
Amazon's threat intelligence team similarly reported an uptick in Artex use, citing campaigns aimed at stealing AI model credentials. Researchers said agentic AI frameworks had "progressed from experimental use to sustained, multi-objective offensive operations that run with minimal human intervention."
While Artex has drawn attention, Google has also observed activity involving CyberStrikeAI, HexStrike AI and PentAGI. Cisco Talos researchers noted these frameworks are lowering the skill threshold for attackers and accelerating intrusions when the tools are misused.
"These frameworks are the new reality," researchers wrote, predicting their evolution will likely include "autonomous lateral movement followed by data identification and exfiltration."