LAMEHUGMalware
- PROMPTSTEAL
According to CERT-UA, LAMEHUG uses an LLM (Qwen) to dynamically generate commands to gather basic information about a computer and recursively exfiltrate Office documents from a set of folders, to be uploaded either by SFTP or HTTP POST requests.From MISP Galaxy
Last reported 1 month ago, by The Hacker News6 reports
ATT&CK techniques
6 of 20 · each with the sentence it comes from- T1053.005Scheduled Task
“It kept a foothold with a scheduled task that beaconed to its command-and-control server every 30 minutes.”
- T1055.012Process Hollowing
“It uses a method called Early Bird APC injection, or process hollowing if the model asks for it.”
- T1071Application Layer Protocol
“T1071”
- T1684Social Engineering
“…rly criminal use focused on leveraging AI tools like ChatGPT to assist in coding malware, generating phishing emails, and crafting social engineering campaigns.”
- T1053Scheduled Task/Job
“Scheduled task/job: cron”
- T1053.003Cron
“T1053.003”
Reports per month
last 12 monthsOctober so far (hatched). Counted by report, not by story.
What the reports say
6 reports, newest first · the exact sentence from each-
This Windows Malware is Built to Let Up to Four AI Models Vote on Its Next Move“LAMEHUG, which Ukraine's CERT-UA reported in July 2025, asked an AI model to write commands for tasks set in its code.”
-
Introducing CAIRN: Frontier tracking for AI-integrated malware“…ts with CAIRN have targeted active malware development since July 2025, when the first AI-integrated samples were reported in the wild (LAMEHUG, CERT-UA).”
-
Fake AI, real malware: Attackers impersonating AI brands“We have not, as of this writing, confirmed a case in our own telemetry, although a prominent public example is LAMEHUG, which CERT-UA attributed with medium confidence to APT28/UAC-0001 (IRON TWILIGHT) and described as using Qwen2.5-Coder-32B-Instruct via Hugging Face to generate…”
-
The Signs Were There: What the First Autonomous Ransomware Case Confirms“LameHug, also tracked as PROMPTSTEAL and reported in July 2025, was the first malware seen querying a live LLM to generate its own commands.”
-
APT28: An Evolution of Tradecraft from X-Agent to LLM Malware“"Кібератаки UAC-0001 на сектор безпеки та оборони із застосуванням програмного засобу LAMEHUG, що використовує LLM (велику мовну модель)"CERT-UA(2025)”
-
Redefining Enterprise Defense in the Era of AI-Led Cyberattacks“Notable cases such as LameHug's (PROMPTSTEAL) use of HuggingFace-hosted AI to craft info-stealing scripts, and how PROMPTFLUX requested obfuscation techniques from Google’s Gemini AI, demonstrate how adversaries are moving past traditional, static malware.”
Indicators in its reports
named in the reports above- 250d4fa37488af9b…90b4f5a90cd7in 1 report
- c4dc171f2513fcaf…422dee5205a7in 1 report
- c13cea04f598e2b0…64e0b983a86fin 1 report
- f5f1f8c3e7b88379…a33adc17a63cin 1 report
- 5191cf625dfc209a…de73c133f3cbin 1 report
- eddbd0ecf7195d38…f130a05a42e5in 1 report
- download-version.1-9-18[.]comin 1 report
- perplexity-ai[.]onlinein 1 report
- code.verification-claude-cdn[.]beerin 1 report
- 45.131.66[.]106in 1 report
- 64.20.53[.]230in 1 report
Copy these 11 indicators · defanged
Thanks — we review every report.