Threat actors are using LLMs and multi-agent frameworks to find and exploit vulnerabilities without human intervention, according to a report released Monday by Google Threat Intelligence Group (GTIG). The findings, which researchers described as a “critical inflection point” in the adversarial AI landscape, underscore a growing asymmetry between machine-speed attacks and human-speed defense, raising concerns that automated adversaries could soon outpace enterprise security teams.
GTIG said it observed for the first time threat actors using LLM frameworks to fully automate vulnerability research and exploitation, with no human in the loop. Between July and September 2026, researchers found an exposed staging server hosting a system called exploit_agent. The multi-agent framework, compatible with multiple frontier and local LLMs, runs continuous differential analysis of patched and unpatched versions of popular web applications. It identifies potential vulnerabilities, drafts exploits and executes them automatically — all without human intervention.
“By chaining these capabilities together, threat actors can dramatically reduce the window between patch disclosure and weaponized exploitation,” GTIG researchers wrote.
They also found scripts that facilitate post-compromise activities, from internal reconnaissance and brute-forcing to installing webshells. Victim telemetry indicated that the operator targeted French websites in the media, entertainment and education sectors.
Previous editions of Google’s quarterly AI threat tracker noted only partial automation of attack workflows.
Enterprise patching windows under pressure
As evidence that threat actors are operationalizing fully automated vulnerability exploitation pipelines, the GTIG researchers cited attacks in August 2026 that exploited vulnerabilities in PaperCut print management servers. According to researchers at GreyNoise, the threat actors apparently used LLMs to develop exploits and target hundreds of organizations in a widespread campaign.
The report reinforces growing concerns about the accelerated exploitation of n-day vulnerabilities, known flaws that often remain unpatched for months or years at a time in enterprise environments. As automated patch diffing and exploit development tools become more sophisticated, defenders can expect to have far less time to remediate vulnerable systems, experts warned.
“Traditional patching cycles are becoming obsolete, and some fixes will need to happen in hours rather than days or weeks,” said Jason Elrod, CISO at MultiCare Health System.
With widespread availability of open source, multi-agent offensive frameworks such as Artex, Cairn, CyberStrikeAI and PentAGI, GTIG warned that adversary tradecraft will continue to evolve. Many security leaders say they need similarly capable tools to stop machine-speed attacks.
“We have to bring AI-enabled defense to bear in order to combat AI-enabled offense,” said Bob Leek, CIO of Clark County, Nev. Automation is one of the county’s top cybersecurity priorities for the coming year, he added.
AI adversaries prioritize speed over stealth — for now
GTIG said today’s agentic frameworks trade stealth for speed, generating heavy network traffic and code-level artifacts that increase the likelihood of detection. But researchers anticipate that will change as AI tools mature.
Threat actors that prioritize stealth and persistence, such as cyberespionage groups, are likely to eschew forensically noisy AI tools. Researchers warned, however, that many financially motivated cybercriminals will sacrifice secrecy for speed and embrace autonomous tooling, “given that the increased reach of their campaigns will likely compensate for any decreases in success rates.”
That strongly suggests pressure on defenders will continue to grow in the months ahead.
"AI cuts attack timelines from days to minutes, so detection-to-containment time has to shrink the same way," said Neil Wyler, vice president of defensive services at cybersecurity services provider Coalfire and senior network operations lead at Black Hat.
Alissa Irei is a TechTarget senior reporter covering cybersecurity news.