Threat Intelligence

Shadow AI Is Hiding From Your Software Inventory: What Bitdefender's New Visibility Tool Looks For

The Hacker News · 13 Oct 2026

Key Takeaway Before blocking AI sites, find out which AI tools, extensions and agents are actually running on your devices, because you cannot manage risk you cannot see.

AI tools are easy to use without leaving a trace in the systems most businesses rely on. According to The Hacker News, a person can paste data into a chat assistant in a browser tab, add a coding assistant as an editor extension, or download a local model runtime that runs models on the machine itself. None of these necessarily show up in a software inventory. As a result, administrators often fall back on blocking well-known AI websites and acting without knowing what is actually running.

Bitdefender's GravityZone AI Visibility and Control (AIVC) aims to answer three questions first: which AI software is on protected Windows endpoints, who is running it, and how risky each item is. Discovery happens during the risk scan, when the GravityZone BEST agent collects metadata about what AI software is present and how it is configured. Bitdefender says it does not inspect prompts, chat history, model weights, credentials, or full file content.

The tool recognises more than 50 AI agents and more than 70,000 skills, which are modular code packages that extend what an agent can do. Each discovered service is placed in one of 17 categories, such as LLMs and coding, so teams can focus on data handling, source code access, or automation risk. The findings appear in the GravityZone Console as two views. This summary is based only on the opening of the article.

Shadow AI Endpoint Security Bitdefender AI Governance

Summarised by CISO AI from The Hacker News, written with Claude Sonnet 5.5. We link back to every original so you can read it yourself.