AI-powered malware families integrating LLMs during execution
Malware Activity
Summary
Hide ▲
Show ▼
Google's GTIG identified multiple AI-powered malware families that use LLMs during execution, signaling a shift toward malware that can adapt while running. The set includes PromptFlux, PromptSteal (LameHug), FruitShell, QuietVault, and PromptLock. The activity spans Windows, macOS, and Linux and raises risk from obfuscation, credential theft, remote C2, and ransomware.
Related Happenings
UAT-9244 TernDoor, PeerTime, and BruteEntry malware activity
Malware Activity
First: 06.03.2026 01:19
Last: 06.03.2026 01:19
Sources 1
About this happening:
A **China-linked** malware cluster has been using **TernDoor**, **PeerTime**, and **BruteEntry** to compromise **telecommunication providers in South America** and turn infected s...
UAT-9244 TernDoor, PeerTime, and BruteEntry malware activity
Malware ActivityAbout this happening: A **China-linked** malware cluster has been using **TernDoor**, **PeerTime**, and **BruteEntry** to compromise **telecommunication providers in South America** and turn infected s...
PromptSpy Android spyware using Google Gemini runtime persistence
Malware Activity
First: 20.02.2026 00:36
Last: 20.02.2026 00:36
Sources 1
About this happening:
**PromptSpy** is an **Android spyware/backdoor** family that uses **Google Gemini** at runtime to help keep the app in the **recent apps** list and automate **Android UI** persist...
PromptSpy Android spyware using Google Gemini runtime persistence
Malware ActivityAbout this happening: **PromptSpy** is an **Android spyware/backdoor** family that uses **Google Gemini** at runtime to help keep the app in the **recent apps** list and automate **Android UI** persist...
Geta RAT, Ares RAT, and DeskRAT cross-platform credential-theft activity
Malware Activity
First: 11.02.2026 16:52
Last: 11.02.2026 16:52
Sources 1
About this happening:
**Geta RAT**, **Ares RAT**, and **DeskRAT** are being deployed across **Windows and Linux** in phishing-led intrusions that enable **credential theft**, **persistent access**, and...
Geta RAT, Ares RAT, and DeskRAT cross-platform credential-theft activity
Malware ActivityAbout this happening: **Geta RAT**, **Ares RAT**, and **DeskRAT** are being deployed across **Windows and Linux** in phishing-led intrusions that enable **credential theft**, **persistent access**, and...
APT36 / SideCopy phishing-led campaign targeting Indian defense organizations
Campaign
First: 11.02.2026 16:52
Last: 11.02.2026 16:52
Sources 1
About this happening:
A **phishing-led** **APT36 / SideCopy** campaign is targeting **Indian defense and government-aligned organizations**, using cross-platform **RATs** to steal sensitive data and ke...
APT36 / SideCopy phishing-led campaign targeting Indian defense organizations
CampaignAbout this happening: A **phishing-led** **APT36 / SideCopy** campaign is targeting **Indian defense and government-aligned organizations**, using cross-platform **RATs** to steal sensitive data and ke...
DEAD#VAX campaign using IPFS-hosted VHD phishing to deploy AsyncRAT
Campaign
First: 04.02.2026 19:24
Last: 04.02.2026 19:24
Sources 1
About this happening:
The **DEAD#VAX** campaign is using **phishing-delivered IPFS-hosted VHD files** to deploy **AsyncRAT**, creating a stealthier path to **fileless endpoint compromise**. The chain r...
DEAD#VAX campaign using IPFS-hosted VHD phishing to deploy AsyncRAT
CampaignAbout this happening: The **DEAD#VAX** campaign is using **phishing-delivered IPFS-hosted VHD files** to deploy **AsyncRAT**, creating a stealthier path to **fileless endpoint compromise**. The chain r...
Timeline
-
05.11.2025 16:59 2 articles · 6mo ago
Google identifies AI-powered malware families using LLMs during execution
Technical Analysis UpdateGoogle Threat Intelligence Group identifies a major shift in 2025 in which adversaries use large language models during execution to support AI-powered malware with "just-in-time" self-modification, including PromptFlux, PromptSteal (a.k.a. LameHug), FruitShell, QuietVault, and PromptLock. The same report also documents Gemini abuse by actors such as APT42, APT41, MuddyCoast (UNC3313), Masan (UNC1069), and Pukchong (UNC4899), along with growing underground-market interest in malicious AI tools; Google disables associated accounts and cuts off PromptFlux access to the Gemini API.
Show sources
- Google warns of new AI-powered malware families deployed in the wild — www.bleepingcomputer.com — 05.11.2025 16:59
- Google warns of new AI-powered malware families deployed in the wild — www.bleepingcomputer.com — 05.11.2025 16:59