Google: attackers use AI agents to weaponize patches and run intrusions
Original title: Google Finds Hackers Using AI Agents to Build and Run Exploits From Software Patches
GTIG says open agent frameworks and LLMs shrink the gap from patch disclosure to n-day exploitation, including PaperCut NG/MF.
Google Threat Intelligence Group told The Hacker News that threat actors have moved from manual prompting to autonomous offensive frameworks such as ARTEX, CAIRN, CyberStrikeAI, and PentAGI. Actors use LLMs for differential analysis of patches and related material to weaponize newly disclosed flaws faster, a pattern GTIG says it saw with PaperCut NG/MF flaws in September. The report also describes stolen LLM credentials resold via APIMart and JetNight, and a Chinese-language phishing platform, Jieyou, that adds an LLM chatbot for lure customization.
Unconfirmed: Alignment of some actors with China is described as suspected, not confirmed.
- Named frameworks include ARTEX, CAIRN, CyberStrikeAI, PentAGI, and Agent Zero.
- ZeroHollow is described as an inference service for exploit development and LLM routing.
- GTIG cites financially motivated actors and suspected China-aligned espionage actors.
- Jieyou lets customers edit phishing panels and templates with natural-language commands.
- Stolen LLM access is resold through underground proxy services APIMart and JetNight.
Impact: Faster n-day weaponization, intrusions, and resale of stolen LLM access
| Product | Affected |
|---|---|
| PaperCut NG/MF | — |
- Organizations
- Google Threat Intelligence Group
- Countries
- China
From the original article
2 min read at thehackernews.com
Google on Monday revealed that adversarial use of artificial intelligence (AI) by threat actors has gone beyond manual prompting to deploying autonomous multi-agent offensive frameworks like ARTEX, CAIRN, CyberStrikeAI, and PentAGI to execute multi-stage intrusions at greater operational velocity. In tandem, threat actors are leveraging the power and capabilities of large language models (LLMs)
thehackernews.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.