PTC News
Search
🔍 Search

Google warns hackers are using AI to attack software without human help

Google says hackers are using an AI tool to find software bugs and launch attacks without human help. Here is what the report says

By  Jasleen Kaur November 17th 2026 06:55 PM -- Updated: November 17th 2026 06:56 PM

PTC Web Desk: Google has warned that hackers are now using AI to find weak spots in software and attack them with almost no human involvement.

The warning comes from the Google Threat Intelligence Group (GTIG) in a new report called “GTIG AI Threat Tracker: Agentic Threat Activity Crosses a Significant Threshold.” Researchers say they found a tool named exploit_agent that handled much of the attack process by itself.

How the tool works

Whenever a company releases a security update, it fixes some bugs. exploit_agent watches for these updates in popular software. It compares the old and new versions of the code, works out which bug was fixed and then plans an attack on organisations that have not yet installed the update.

Also Read | Is your iPhone safe? Optimizely supply-chain attack pushed iOS exploit kits via 178 organisations sites

GTIG found the tool on an exposed staging server. It was linked to several large language models, including both leading commercial models and ones hosted locally.

Why old bugs are an easy target

Google says attackers may prefer going after bugs that already have a fix (called n-days) instead of hunting for brand-new, unknown flaws (zero-days). Once a patch is announced, hackers get plenty to work with, such as the code changes, release notes, research papers and sample exploit code.

What else was on the server

The same server held data stolen from victims. It also had scripts for scanning internal networks, brute-force attacks, installing webshells, and removing webshells planted by rival hackers. Victim data suggested the operator was mostly interested in French websites in the media, entertainment and education sectors. A comment in Russian in the configuration file told the operator to enter AI API keys “right here, not in the chat.”

PaperCut example

The report points to two flaws in PaperCut print management servers, which were exploited as zero-days in late August 2026. Researchers and the security firm Huntress built working exploits using AI within a day of the disclosure. Four days later, GreyNoise spotted a campaign that likely used AI to build and test exploits before hitting hundreds of organisations. In one case, an attacker went from first entry to full domain administrator access within minutes.

Speed is the real problem

Google says hackers have not invented a new technique. They are using AI to move faster and at a larger scale. “For the most part, attackers aren’t using new groundbreaking methods to break into systems — they are still going after known, unfixed issues, just with much greater speed and scale,” said John Hultquist, chief analyst at GTIG.

He added that companies must adopt AI-based defence too. “Instead of relying on manual effort to keep up, we will have to embrace agentic defence. Or we risk falling behind,” he said.

Google says it is using its own AI models and working with defenders to find and fix flaws before attackers can use them.

Growing concern

The report arrives as AI-linked attacks rise. One recent case was the $6.3 billion Binance breach, where hackers abused an internal AI tool of the crypto platform to gain wider access. Last month, Microsoft said a China-linked group used an AI framework called ARTEX to launch automated attacks on hundreds of organisations worldwide. India’s CERT-In has also earlier warned that hackers are using AI tools to target organisations.

What you can do

Install software updates as soon as they are released. With AI now speeding up attacks, even a few days of delay can leave a system open.

Related Post