Journalism begins where hype ends

,,

The rise of powerful AI will be either the best, or the worst thing, ever to happen to humanity. We do not yet know which."

—Stephen Hawking

“A Hacker That Never Sleeps”: Google Flags Autonomous AI Attack Frameworks

Google’s latest threat intelligence report says attackers have moved beyond basic prompting to autonomous multi-agent frameworks and LLMjacking.
Laptop showing a red warning symbol, illustrating Google’s AI threat intelligence report on autonomous AI attacks
November 17, 2026 07:25 AM IST | Written by Vaibhav Jha

A latest artificial intelligence (AI) threat intelligence report from Google says that threat actors have moved beyond “basic-prompting” AI models and begun deploying autonomous multi-agent frameworks and using sophisticated techniques like “LLMjacking”.

The increasingly diverse use of AI helps the threat actors to “compress attack timelines, scale operations and execute multi-stage intrusions at machine speed.”

The report also mentions how LLMs are being used by state-sponsored groups and commercial Influence-as-a-Service (IaaS) vendors to produce synthetic media at scale, create localized personas and automate cross-platform distribution.

“Imagine a hacker that doesn’t eat, sleep, or get tired, and never stops searching for a way in. Bad actors are now using autonomous LLM frameworks to continuously scan software updates, figure out what flaw was patched, and build an attack for it — all without human involvement,” read a report from Google Threat Intelligence Group (GTIG).

According to the latest threat intelligence report prepared by GTIG and Mandiant, the key trends observed in the last quarter shows a clear shift to autonomous workflows and multi-agent frameworks as adversaries deploy CAIRN, an open source, multi-agent pentesting framework, Hexstrike AI, an open source offensive Model Context Protocol (MCP)-integrated harness , CyberStrike AI, an open source, LLM-native offensive framework and PentAGI, an open source autonomous offensive framework to execute intrusions at greater speed.

A latest report from Unit 42 of Palo Alto says that unknown attackers allegedly speaking Chinese, had used CAIRN, an open-source AI penetration testing framework to target at least 19 organizations autonomously. The attackers had paired CAIRN with AI models DeepSeek and Zhipu GLM.
Similarly, the GTIG-Mandiant threat report observes that threat actors are now using large language models (LLMs) to fully automate the reverse engineering of patches via differential analysis and also conduct LLMJacking.

What is LLMJacking?

LLMJacking, as the name suggests, involves threat actors “hijacking” enterprise access to LLMs and then generate large spikes in fraudulent cloud charges in a matter of hours, at times even exceeding $100,000 USD in a single day. A latest November 2026 report from Mandiant shows that LLMJacking impacted organizations in North America, South America and Asia Pacific. Similarly the GTIG threat intelligence report November 2026 observes use of LLMs for target research, automated data analysis and novel supply chain attack vectors.

According to the report by Google, the threat actors’ footprints allegedly shows a People’s Republic of China (PRC) nexus.“We have observed several threat actors separately experimenting with the use of these frameworks and harnesses, including financially motivated, as well as suspected espionage threat actors with a PRC nexus. This includes evidence that a PRC-nexus group was developing its own automated attack framework for likely use in cyber espionage operations,” read the report.

The report also states that due to unrestricted autonomous executions in these cyber attacks, there is forensic evidence of distinct code-level traces making it possible to recognize patterns.

Also Read: Study Identifies How AI Safeguard Failures Could Enable a Major Terrorist Attack

Author

  • Vaibhav Jha, editor and co-founder at AI FrontPage

    Vaibhav Jha is an Editor and Co-founder of AI FrontPage. In his decade long career in journalism, Vaibhav has reported for publications including The Indian Express, Hindustan Times, and The New York Times, covering the intersection of technology, policy, and society. Outside work, he’s usually trying to persuade people to watch Anurag Kashyap films.

    LinkedIn