Google finds attacker AI framework that finds and runs exploits
Original title: Google spots the first hacker AI system that finds and exploits flaws autonomously
GTIG and Mandiant say exploit_agent targets n-days and was found with victim data; theft attribution is unproven.
Google Threat Intelligence Group and Mandiant reported a hacker-operated multi-agent framework, exploit_agent, found on an exposed staging server. It connects to multiple LLMs, reviews code changes after patches, identifies vulnerabilities, generates exploits, and can run them without step-by-step operator direction. The server also held exfiltrated victim data and post-compromise scripts, with interest in French media, entertainment, and education sites. Google did not establish that autonomous operation caused the theft. Separate observations covered likely AI-assisted PaperCut exploit development, CAIRN intrusions including US aerospace, and a PentAGI-based platform scanning organizations in Asia.
Unconfirmed: Google did not link the observed theft to autonomous exploit_agent activity, and PaperCut AI use is described as likely.
- GreyNoise research cited a PaperCut campaign four days after disclosure against hundreds of organizations.
- CAIRN was used in intrusion activity that included the US aerospace sector.
- A customized PentAGI platform performed reconnaissance and vulnerability scanning in Asia.
- Analysts said the techniques are accelerated existing attacks and often create detectable activity.
Impact: Automated discovery and execution of n-day exploits, plus observed data theft
| Product | Affected |
|---|---|
| PaperCut print management servers | — |
- 1.Fast-track patches and consider virtual patching for internet-facing systems.
- 2.Use MFA, strong passwords, and controls against repeated logins.
- Vendors
- PaperCut
- Malware
- exploit_agentCAIRNPentAGI
- Organizations
- Google Threat Intelligence GroupMandiantGreyNoise
- Countries
- FranceUnited States
From the original article
4 min read at csoonline.com
Google said it had identified a hacker-operated AI framework designed to discover software vulnerabilities and execute exploits without human intervention, marking the first time its researchers had observed such a system being used by threat actors.
csoonline.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.