INSPECTING ARCHIVED INTELLIGENCE (OUTDATED VERSION).
PoeLLM malware uses poem to control AI infrastructure
|
2026-10-08 09:56
LOW
MEDIUM
AI-ENABLED ATTACK · AUTONOMOUS
MALWARE & BOTNETS
Executive Summary
AI-generated
Hackers used a poem hosted on GitHub to control PoeLLM malware targeting AI infrastructure, turning exposed systems into a growing cryptocurrency-mining botnet. The campaign affected over 3,000 servers primarily located in the US and Western Europe, including LiteLLM and Ollama services, as well as Gotenberg PDF converters and Gitea development servers. The attackers used an adversarial poetry technique to trick LLMs into bypassing safety guardrails, hiding malicious messages within seemingly innocuous poems that could not be easily identified by security researchers without access to the malware referencing them.
Technical Mitigations AI-generated
• Apply the vendor fix for CVE-2026-10520 and treat internet-facing systems that were not patched in time as potentially compromised until verified.
Technical Observables 23 IoCs Login Required
Indicators of Compromise (IoCs) Protected
Raw threat telemetry, malicious IP addresses, file hashes, and direct VirusTotal correlation are restricted to authenticated users.
files 1
ips 16
urls 3
sha256 1
domains 2
Intelligence Metadata
Actors / Malware / CVEs / Campaigns
CVE-2026-10520CVE-2026-10520
Target & Sectors
EUROPE
EUROPE
technologytechnology
Incident Timeline
at least April 2026
Threat actors used a GitHub-hosted poem to control the PoeLLM malware targeting AI infrastructure.
Click on any entity below to view its context and source!
tactic
T1588.001 - Malware
Active since at least April 2026, PoeLLM uses selected words in […] The post Hackers Use GitHub-Hosted Poem to Control PoeLLM Malware Targeting AI Infrastructure appeared first on Cyber Security News .
organisation
Cyber Security News
Active since at least April 2026, PoeLLM uses selected words in […] The post Hackers Use GitHub-Hosted Poem to Control PoeLLM Malware Targeting AI Infrastructure appeared first on Cyber Security News .
2026/10/08
Threat actors used an adversarial poem hosted on GitHub to control the PoeLLM malware, targeting AI infrastructure and assembling a sweeping botnet that has infected over 3,400 servers since April.
Click on any entity below to view its context and source!
infrastructure
Ivanti
In addition to these open source tools, the attacker may have targeted commercial software including Ivanti Sentry.
The threat hunters first spotted the PoeLLM malware while investigating an Ivanti Sentry vulnerability, CVE-2026-10520.
“Shortly after contacting this C2, the Ivanti Sentry victim began scanning for other vulnerable devices.”
organisation
Gitea
That discovery uncovered a sweeping exploit-scanning and cryptocurrency-mining botnet linked to multiple compromised services and tools, including LiteLLM, Ollama, Gotenberg and Gitea.
The campaign targets internet-facing services, including LiteLLM and Ollama, while also affecting Gotenberg PDF converters and Gitea development servers.
Additionally, hundreds of victims were running Gotenberg, a PDF converter, and software development platform Gitea.
organisation
GitHub
Hackers are using a poem hosted on GitHub to guide PoeLLM malware toward its command-and-control servers, turning exposed AI infrastructure into a growing cryptocurrency-mining botnet.
When the malware retrieves the updated poem from GitHub, it extracts four specific words based on their positions relative to fixed text anchors within the verse.
infrastructure
3,000 servers
A suspected Italian attacker armed with a malware-controlling poem has infected more than 3,000 servers since April, breaking into enterprise AI infrastructure to mine cryptocurrency and add compromised systems to its growing botnet.
There would be no reason for any security researcher to identify this poem as malicious - or know about the IP address hidden within it - unless they had access to the malware referencing it.”
PoeLLM malware has been active since at least April, impacting more than 3,000 servers primarily located in the US and Western Europe, and it continues to infect new victims.
organisation
Ollama
The campaign targets internet-facing services, including LiteLLM and Ollama, while also affecting Gotenberg PDF converters and Gitea development servers.
“AI makes it easier to deploy tools like LiteLLM, Ollama, or Gotenberg, but AI isn't always checking to make sure those services are patched and protected from attackers,” the researchers said.
organisation
IP
There would be no reason for any security researcher to identify this poem as malicious - or know about the IP address hidden within it - unless they had access to the malware referencing it.”
PoeLLM malware has been active since at least April, impacting more than 3,000 servers primarily located in the US and Western Europe, and it continues to infect new victims.
“The most interesting piece of the poem approach is that the IP address used for C2 communications is invisible outside of the victim’s netflow.
organisation
Black Lotus Labs
How adversarial poetry works
Black Lotus Labs attributed the PoeLLM malware to an Italian-speaking criminal, and named the financially motivated campaign Canto Incognito because it hides the malicious commands in a poem posted to a GitHub repository.
In other malware samples, there might be a hard-coded fallback C2 address, or URL which points to an IP address written elsewhere on the internet,” Ryan English, information security engineer at Black Lotus Labs, told CyberScoop.
organisation
Lumen Technologies’ Black Lotus Labs
Malware that takes technical cues from a poem to assemble a growing botnet by targeting open-source AI services has compromised more than 3,400 servers since April, Lumen Technologies’ Black Lotus Labs said in a report Wednesday.
infrastructure
3,400 servers
Malware that takes technical cues from a poem to assemble a growing botnet by targeting open-source AI services has compromised more than 3,400 servers since April, Lumen Technologies’ Black Lotus Labs said in a report Wednesday.
organisation
Gotenberg PDF
The campaign targets internet-facing services, including LiteLLM and Ollama, while also affecting Gotenberg PDF converters and Gitea development servers.
organisation
PDF
Additionally, hundreds of victims were running Gotenberg, a PDF converter, and software development platform Gitea.
organisation
Ivanti Sentry
In addition to these open source tools, the attacker may have targeted commercial software including Ivanti Sentry.
organisation
Lumen’s Black Lotus Labs
This is the first case of “
adversarial poetry
” - an AI jailbreak technique that turns harmful prompts into poems to trick LLMs into bypassing safety guardrails - that Lumen’s Black Lotus Labs, which has been tracking the PoeLLM malware, has seen in real-world attacks.
organisation
CyberScoop
In other malware samples, there might be a hard-coded fallback C2 address, or URL which points to an IP address written elsewhere on the internet,” Ryan English, information security engineer at Black Lotus Labs, told CyberScoop.
organisation
Kryptex
The malware deploys XMRig and Iron miners, and connects victims to Kryptex mining infrastructure.
organisation
GPU
In addition to using compromised GPU hardware powering AI workloads to mine cryptocurrency, PoeLLM also turns victims' machines into vulnerability scanners and exploit servers, which allows the attacker to compromise even more vulnerable systems.
organisation
CI
For comparison: The
LiteLLM supply chain attack
, which began with a
compromised Trivy build
, potentially exposed
more than 2,500 companies
and 434,000 CI/CD pipelines worldwide, according to CloudSEK security researchers.
infrastructure
434,000 CD pipelines
For comparison: The
LiteLLM supply chain attack
, which began with a
compromised Trivy build
, potentially exposed
more than 2,500 companies
and 434,000 CI/CD pipelines worldwide, according to CloudSEK security researchers.
infrastructure
800 active servers
At its peak, the malware infected more than 800 active servers per day.
data_breach
4 bytes
…nd " threading"
Word 3: text between "Beyond the wall of " and ","
0x44a8db–0x44a99b extracts the fourth word differently:
Find " of distant servers"
Walk backward to the previous whitespace
Require the 4 bytes before the word to be "the "
Use the word after "the " as Word 4
The four words are then matched to corresponding numbers, which combine to form the IPv4 address hosting the server.
victims
2,500 victims
“Other notable campaigns this year, including the
LiteLLM supply chain compromise
, focused on a single service and impacted roughly 2,500 victims, according to open sources.
victims
3,000 PoeLLM victims
The collection of more than 3,000 PoeLLM victims appears to exhibit multiple vulnerable services at any given time.”
organisation
C2s
“Many of the C2s used in this campaign were never detected on crowd-sourced security tools, indicating that this layer of obfuscation was very helpful in improving the resilience of a C2,” English said.
Tactical Metrics
Metrics
infrastructure
3,000
Servers
Click for context!
There would be no reason for any security researcher to identify this poem as malicious - or know about the IP address hidden within it - unless they had access to the malware referencing it.”
PoeLLM malware has been active since at least April, impacting more than 3,000 servers primarily located in the US and Western Europe, and it continues to infect new victims.
A suspected Italian attacker armed with a malware-controlling poem has infected more than 3,000 servers since April, breaking into enterprise AI infrastructure to mine cryptocurrency and add compromised systems to its growing botnet.
Metrics
infrastructure
Ivanti
Affected Product
In addition to these open source tools, the attacker may have targeted commercial software including Ivanti Sentry.
The threat hunters first spotted the PoeLLM malware while investigating an Ivanti Sentry vulnerability, CVE-2026-10520.
“In early June 2026, a compromised Ivanti Sentry victim contacted a dedicated server at 5.78.73[.]122,” according to a Wednesday
report
shared with
The Register
.
“Shortly after contacting this C2, the Ivanti Sentry victim began scanning for other vulnerable devices.”
Researchers first encountered PoeLLM infrastructure in June during an investigation into a maximum-severity defect affecting Ivanti’s secure mobile gateway product, Sentry.
Metrics
infrastructure
434,000
Cd Pipelines
For comparison: The
LiteLLM supply chain attack
, which began with a
compromised Trivy build
, potentially exposed
more than 2,500 companies
and 434,000 CI/CD pipelines worldwide, according to CloudSEK security researchers.
Metrics
infrastructure
800
Active Servers
At its peak, the malware infected more than 800 active servers per day.
Metrics
data_breach
4
Bytes
…nd " threading"
Word 3: text between "Beyond the wall of " and ","
0x44a8db–0x44a99b extracts the fourth word differently:
Find " of distant servers"
Walk backward to the previous whitespace
Require the 4 bytes before the word to be "the "
Use the word after "the " as Word 4
The four words are then matched to corresponding numbers, which combine to form the IPv4 address hosting the server.
Metrics
victims
2,500
Victims
“Other notable campaigns this year, including the
LiteLLM supply chain compromise
, focused on a single service and impacted roughly 2,500 victims, according to open sources.
Metrics
victims
3,000
Poellm Victims
The collection of more than 3,000 PoeLLM victims appears to exhibit multiple vulnerable services at any given time.”
Metrics
infrastructure
3,400
Servers
Malware that takes technical cues from a poem to assemble a growing botnet by targeting open-source AI services has compromised more than 3,400 servers since April, Lumen Technologies’ Black Lotus Labs said in a report Wednesday.
Intelligence Sources
The Register - CSO
2026-10-07
Poetry is the new AI security threat as PoeLLM malware infects 3K+ servers
The Register - CSO
AlienVault OTX
2026-10-08
CyberScoop
2026-10-07
Unpublish from Social Media?
Are you sure you want to delete this podcast video from all synchronized social networks (YouTube, Facebook, Threads)?
Important:
Due to Meta API restrictions, Instagram Reels cannot be deleted automatically via API by third-party apps.
View Profile to Delete Manually
View Profile to Delete Manually
Tactical Intelligence
Report Intelligence Issue
Podcast Options
Generate
Incident Version History
CURRENT VERSION
Last Updated: 2026-11-12T15:41
Comprehensive Tactical Telemetry
Highly Correlated Entities
17x
organisation
Identified Entity
GitHub
entity
4x
timeline
Temporal Reference
at least April 2026
date
4x
general metric
Word
1
word
3x
tactic
Cyber Operation Type
Botnet
tactic
2x
target region
Target Country
United States
country
2x
infrastructure
Servers
3,000
servers
Contextual Telemetry
Context Block
16 METRICS
industry
Targeted Sector
Technology
sector
tactic
MITRE ATT&CK Technique
T1588.001 - Malware
technique
target region
Target Region
EUROPE
region
infrastructure
Affected Product
Ivanti
software
vulnerability
Exploited CVE
CVE-2026-10520
cve
attribution
Attributing Entity
CloudSEK
authority
general metric
Companies
2,500
companies
infrastructure
Cd Pipelines
434,000
cd pipelines
infrastructure
Active Servers
800
active servers
general metric
Times
11
times
data breach
Bytes
4
bytes
victims
Victims
2,500
victims
victims
Poellm Victims
3,000
poellm victims
general metric
Driver
92
driver
general metric
Diode
119
diode
general metric
String
74
string
Click on any entity below to view its context in the main text!
Selective Unpublish
Select the networks you want to remove this post from. The system will try to delete the real post through the API and clean the database so you can publish it again.
By navigating this website, you accept the use of strictly necessary technical cookies for session security and basic platform functionality. We do not use tracking or advertising cookies. Read our Privacy Policy.
Generating report…
0s