Economy

Self-executing AI hacking tool with Russian comments discovered in Google archives – Check your exposure

By Clayton · November 18, 2026

Google’s security team has identified a self-executing AI tool that autonomously discovers and exploits software vulnerabilities without any human intervention, and the code was found stored in Google’s own public archives with comments written in Russian — raising the stakes for every organization that relies on open-source dependencies.

What the AI tool does that human hackers cannot

The tool, which Google discovered in its own archives, is not a theoretical proof-of-concept. It is a working piece of code that can scan a target system, identify a flaw, and then execute an exploit — all without a person pressing a button.

Most automated security tools still require a human to validate findings or write the exploit. This one cuts out that step entirely, meaning the time between a vulnerability being discovered and an attack can drop to near zero.

Why the Russian comments matter

The code’s comments are written in Russian, according to Google’s analysis. That detail does not prove the tool originated in Russia — comments in any language can be faked — but it does narrow the pool of likely developers and suggests the tool was not built for public, open-source collaboration.

Google has not named any specific group or state actor behind the code, and the company has not said whether the tool was ever used in a live attack before it was found.

How to check your exposure

Because the tool was stored in Google’s public archives, it is possible that copies were downloaded before Google took action. The risk is highest for organizations that use software libraries or frameworks that the tool is known to target.

  1. Review your dependency list — check every open-source library your applications pull from, especially those with recent vulnerability disclosures.
  2. Run a scan of your public-facing systems for any of the specific exploit signatures Google has flagged internally (Google has not published the full list publicly).
  3. Monitor your logs for unusual automated traffic that attempts to probe multiple endpoints in rapid succession — that pattern matches the tool’s behavior.
  4. Patch any known critical vulnerabilities within 24 hours, because the tool can exploit them the moment they are disclosed.

What happens next

Google has not said whether it will release the tool’s code or a detailed technical breakdown to the security community. That decision will determine whether defenders can build detection rules or whether the tool remains a black-box threat.

Security teams should treat this as a warning shot: the era of fully autonomous exploit tools has arrived, and the only defense is faster patching and better monitoring — because the AI does not sleep.

Leave a Reply

Your email address will not be published. Required fields are marked *