Google Discovers Autonomous AI Hacking Tool With Russian Code Comments

Google Discovers Autonomous AI Hacking Tool With Russian Code Comments

Google identifies first fully automated AI attack tool

Google has discovered a new hacking tool that uses artificial intelligence to find and exploit software flaws automatically. The tool, which researchers named "exploit_agent," can operate entirely without human help. This is the first time Google has observed attackers using AI frameworks to fully automate the discovery and use of software vulnerabilities.

The tool was found on an exposed staging server, which is a computer used by attackers to store and prepare their digital weapons. Google researchers noted that the tool was connected to several large language models (LLMs). These are the AI systems that power chatbots by processing and generating human-like text.

How the autonomous tool targets software flaws

  • The tool monitors web software for new security patches, which are updates released to fix errors.
  • It compares the software code before and after a fix to identify exactly where the weak point was located.
  • It automatically drafts an "exploit," or a piece of code designed to take advantage of that weakness.
  • The AI then runs the exploit code on its own to break into systems.

Official findings from Google Threat Intelligence

The Google Threat Intelligence Group (GTIG) detailed these findings in its third-quarter report. The report explained that the tool could guess passwords by trying many combinations, a method known as brute-forcing. It also included scripts to map out the inside of a network after an initial break-in and could plant hidden control panels, called webshells, to allow attackers to return later.

Files associated with the tool contained comments written in Russian. One note translated by Google instructed the operator to enter API keys, which act as passwords for AI services, directly into the code rather than in a chat interface. Evidence on the server suggested the tool was used to target French websites in the media, education, and entertainment sectors.

Confirmed attacks on cryptocurrency firms

Google confirmed that it has also linked North Korean hackers to separate AI-powered attacks specifically targeting cryptocurrency companies. Cryptocurrency is a digital currency that uses a decentralized system to record transactions. These hackers are reportedly using AI to make their attacks more effective.

Why autonomous AI agents pose a risk

The discovery of self-running hacking tools adds to growing concerns about AI agents. Federal Reserve Governor Michael Barr recently warned that autonomous AI could trigger a bank run in a matter of minutes. Other financial experts have noted that AI assistants could drain bank deposits by automatically moving money to higher-paying accounts faster than banks can react.

Sources

Newisty Editorial Team
Written by

Newisty Editorial Team

Technology · Crypto · Digital Economy
View all posts

Newisty Editorial Team covers technology, cryptocurrency, digital products, online platforms, developer tools and the wider digital economy. Our content is researched from official sources, company announcements, public documentation, market data and other primary or reputable sources. Articles are reviewed and edited before publication for clarity, accuracy and useful context.

Comments (0)

Leave a comment
Your comment will appear publicly after submission.
No comments yet. Be the first to comment!