VulnHub

Real-time Cybersecurity News

Researchers Find 175,000 Publicly Exposed Ollama AI Servers Across 130 Countries

The Hacker News
SentinelOne

Overview

A recent investigation by SentinelOne SentinelLABS and Censys uncovered 175,000 publicly accessible Ollama AI servers spread across 130 countries. These servers, which are part of an open-source AI deployment, are found in both cloud environments and residential networks. The exposure of these systems poses significant security risks as they operate without proper management or oversight. This unmanaged infrastructure could be exploited by malicious actors for various purposes, including data breaches or launching attacks. Companies and users relying on these servers should take immediate action to secure their systems and limit exposure to potential threats.

Key Takeaways

  • Affected Systems: Ollama AI servers
  • Action Required: Organizations should secure their AI servers by implementing access controls, monitoring network traffic, and ensuring proper configurations are in place to limit exposure.
  • Timeline: Newly disclosed

Original Article Summary

A new joint investigation by SentinelOne SentinelLABS, and Censys has revealed that the open-source artificial intelligence (AI) deployment has created a vast "unmanaged, publicly accessible layer of AI compute infrastructure" that spans 175,000 unique Ollama hosts across 130 countries. These systems, which span both cloud and residential networks across the world, operate outside the

Impact

Ollama AI servers

Exploitation Status

The exploitation status is currently unknown. Monitor vendor advisories and security bulletins for updates.

Timeline

Newly disclosed

Remediation

Organizations should secure their AI servers by implementing access controls, monitoring network traffic, and ensuring proper configurations are in place to limit exposure.

Additional Information

This threat intelligence is aggregated from trusted cybersecurity sources. For the most up-to-date information, technical details, and official vendor guidance, please refer to the original article linked below.

Related Topics: This incident relates to SentinelOne.

Read Original Article

Related Coverage

Owner of Incognito dark web drugs market gets 30 years in prison

BleepingComputer

A Taiwanese man has been sentenced to 30 years in prison for running Incognito Market, a major dark web platform that facilitated the sale of over $105 million in illegal drugs globally. The market operated for several years, connecting buyers and sellers in a largely anonymous online environment. Authorities have emphasized the significance of this case in combating the proliferation of illegal drug trade online. The sentencing serves as a stern warning to others involved in similar activities, highlighting the ongoing efforts to dismantle dark web marketplaces that contribute to the trafficking of narcotics. This incident underscores the challenges law enforcement faces in regulating online criminal activities and the need for continued vigilance in cybersecurity measures.

Feb 4, 2026

Google Looker Bugs Allow Cross-Tenant RCE, Data Exfil

darkreading

Recent vulnerabilities in Google Looker have raised serious concerns about security, particularly regarding cross-tenant remote code execution (RCE) and data exfiltration. Attackers could exploit these flaws to gain access to environments of other Google Cloud Platform (GCP) tenants by leveraging a compromised Looker user account. This means that sensitive data from multiple organizations could potentially be at risk, making it a significant threat for businesses relying on GCP services. The findings underscore the need for users and companies to review their security practices and ensure that they are protected against unauthorized access. As vulnerabilities like these can lead to major data breaches, prompt action is essential to safeguard sensitive information.

Feb 4, 2026

The First 90 Seconds: How Early Decisions Shape Incident Response Investigations

The Hacker News

The article discusses how the initial moments of an incident response can significantly impact the outcome of an investigation. It emphasizes that many failures in incident response are not due to a lack of tools or expertise but rather the decisions made immediately after detecting an incident. High-pressure situations and incomplete information can lead teams to lose control over their investigations, even when they have the capability to manage the intrusion effectively. The author shares experiences of both successful recoveries and failures, underscoring the need for clear protocols and calm decision-making during the critical first 90 seconds after an incident is detected. This insight is essential for organizations looking to improve their incident response processes.

Feb 4, 2026

Avast brings deepfake scam detection to Windows PCs and mobile devices

Help Net Security

Avast has rolled out two new security features aimed at detecting scams involving deepfake technology. The Avast Scam Guardian and Scam Guardian Pro are now available for mobile devices, while the Avast Deepfake Guard is launched for Windows PCs. This AI-driven tool is designed to analyze and identify harmful audio that could be embedded within video content. The goal is to enhance protection for users against various scam tactics that can occur through text messages, phone calls, and video platforms. These updates are particularly relevant as the rise of deepfakes poses a growing risk to online security, making it essential for users to have reliable tools to safeguard against such threats.

Feb 4, 2026

Security Analysis of Moltbook Agent Network: Bot-to-Bot Prompt Injection and Data Leaks

SecurityWeek

Wiz and Permiso have discovered significant security vulnerabilities in the Moltbook Agent Network, which is an AI agent social network. Their analysis reveals that bot-to-bot prompt injection attacks could allow malicious bots to manipulate other bots, leading to unauthorized actions or data leaks. This poses a risk to users relying on these AI agents for various tasks, as sensitive information could be compromised. The findings indicate that these vulnerabilities could be exploited by attackers to gain control over the network and access confidential data. As AI technologies become more prevalent, it is crucial for developers to address these security flaws to protect users and maintain trust in AI systems.

Feb 4, 2026

Global Threat Map: Open-source real-time situational awareness platform

Help Net Security

The Global Threat Map is an open-source initiative designed to provide security teams with real-time visibility of cyber incidents worldwide. It aggregates various open data feeds into an interactive map that displays key indicators like malware spread, phishing attempts, and attack traffic based on geographic location. Unlike traditional threat maps, which are often produced by security vendors, this project relies on community contributions to maintain and update the data. This platform is particularly valuable for organizations looking to enhance their situational awareness and respond to emerging threats more effectively. By utilizing open-source data, it fosters collaboration among security professionals and helps them stay informed about the latest cyber activities that could impact their operations.

Feb 4, 2026