SShortSingh.
Back to feed

Cryptomining botnet PoeLLM targets AI infrastructure servers, researchers warn

0
·4 views

A cryptomining botnet called PoeLLM is targeting servers running popular AI and development services, according to Lumen's Black Lotus Labs. The malware uses a poem on GitHub to coordinate its command-and-control infrastructure by translating words into IP addresses. Researchers identified compromised deployments running LiteLLM, Ollama, Gotenberg, and Gitea, with activity traced back to April 2026. The botnet uses infected hosts to mine cryptocurrency and scan for additional victims. Security researchers advise administrators of these services to verify their network configurations and update vulnerable software.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Drone aerial surveys reveal critical security blind spots in perimeter assessments

Traditional ground-level perimeter security assessments often miss vulnerabilities that are only visible from above. Using drones, certified pilots can efficiently inspect camera sightlines, mounting hardware, and fence continuity from an aerial perspective. These flights provide documented, timestamped evidence of issues like corrosion, blind spots, and physical breaches that are invisible from the ground. For industrial sites, such drone operations require proper certification and regulatory compliance. Incorporating an aerial pass into security audits reveals critical findings that ground surveys alone cannot detect.

0
ProgrammingDEV Community ·

OpenAI's annualized revenue projection is $20 billion lower than previously indicated.

OpenAI has reportedly revised its annualized revenue projections downward by $20 billion. This adjustment reduces the figure from a previously signaled, higher expectation. The company, known for developing ChatGPT, is a major player in the artificial intelligence sector. The change highlights the volatility and challenges in forecasting within the high-growth tech industry.

0
ProgrammingDEV Community ·

Open-source AI virtual try-on tool aims to reduce fashion returns and screen time

A developer has built a virtual try-on web application for the Hacktoberfest Open-Source AI Challenge. The app allows users to upload their own photo and a garment image to see a simulated outfit result. It leverages Google's open-weight Gemma AI model to analyze garments and suggest fit. The tool is designed for college students and online shoppers seeking to minimize returns and reduce time spent shopping online. A live demo and source code for the project have been made publicly available.

0
ProgrammingDEV Community ·

Jev vs Gemini Flash-Lite evaluation shows trade-offs between cost and accuracy

The author compared the Jev decision model and Google's Gemini Flash-Lite LLM for a classification task in October 2026. They tested both models thousands of times on a production task to route system checks. Jev is designed to output calibrated probabilities over options at a very low cost. Gemini Flash-Lite is a low-cost model optimized for high-volume tasks like classification. The test aimed to determine the most cost-effective and reliable setup for their routing system.