Cyber Security News

GitLab Vulnerabilities Expose Users to Prompt Injection Attacks and Data Theft

GitLab has released critical security patches addressing nine vulnerabilities across Community Edition (CE) and Enterprise Edition (EE), including a concerning prompt injection flaw in GitLab Duo that could expose sensitive information from confidential issues.

The company is urging all self-managed installations to upgrade immediately to versions 18.5.2, 18.4.4, or 18.3.6.

The most alarming vulnerability is CVE-2025-6945, a prompt injection flaw in GitLab Duo’s review feature that allows authenticated users to leak sensitive information from confidential issues by injecting hidden prompts into merge request comments.

CVE IDVulnerability TitleSeverityCVSS Score
CVE-2025-11224Cross-site scripting in k8s proxyHigh7.7
CVE-2025-11865Incorrect authorization in workflowsMedium6.5
CVE-2025-2615Information disclosure in GraphQL subscriptionsMedium4.3
CVE-2025-7000Information disclosure in access controlMedium4.3
CVE-2025-6945Prompt injection in GitLab Duo reviewLow3.5
CVE-2025-6171Information disclosure in packages APILow3.1
CVE-2025-11990Client-side path traversal in branch namesLow3.1
CVE-2025-7736Improper access control in GitLab PagesLow3.1
CVE-2025-12983Denial of service in markdownLow3.1

This attack demonstrates how AI-powered features can become security risks when input validation fails.

The patch batch also includes a high-severity cross-site scripting vulnerability (CVE-2025-11224) in the Kubernetes proxy functionality, which could allow authenticated users to execute stored XSS attacks due to improper input validation.

This affects GitLab versions back to 15.10, creating a significant exposure window for organizations running older instances.

Additionally, GitLab addressed two medium-severity information disclosure issues that could grant unauthorized access to sensitive data.

CVE-2025-2615 allows blocked users to access confidential information through GraphQL WebSocket subscriptions.

At the same time, CVE-2025-7000 permits unauthorized users to view confidential branch names by accessing project issues with related merge requests. These flaws highlight gaps in GitLab’s access control mechanisms.

Enterprise Edition users should pay attention to CVE-2025-11865. This medium-severity authorization bypass allows users to remove another user’s Duo workflows.

This vulnerability underscores the need for stricter permission validation in workflow management systems.

The remaining six vulnerabilities carry lower severity ratings but still warrant attention. CVE-2025-6171 enables authenticated reporters to view restricted branch names and pipeline details through the packages API endpoint.

CVE-2025-7736 allows users to bypass access controls and access GitLab Pages content through OAuth provider authentication.

CVE-2025-11990 presents a client-side path-traversal risk via repository references and redirect-handling weaknesses.

At the same time, CVE-2025-12983 can trigger denial-of-service conditions via specially crafted Markdown with nested formatting patterns.

GitLab recommends immediate action for all affected installations. GitLab.com users are already running patched versions, and Dedicated customers require no action.

Security researchers participating in HackerOne’s bug bounty program reported most of the vulnerabilities, demonstrating the value of coordinated disclosure.

The company also updated libxslt to version 1.1.43, patching additional security issues, including CVE-2024-55549 and CVE-2025-24855.

Follow us on Google News, LinkedIn, and X to Get Instant Updates and set GBH as a Preferred Source in Google.

Divya

Divya is a Senior Journalist at GBhackers covering Cyber Attacks, Threats, Breaches, Vulnerabilities and other happenings in the cyber world.

Recent Posts

Hackers Use ConnectWise SkyKick Credentials to Access Microsoft 365, Halting Cloud Backups

Attackers abused compromised ConnectWise Cloud Backup credentials to access Microsoft 365 customer environments, prompting ConnectWise…

1 day ago

Anthropic’s Claude AI Uncovers Origins of NSA-Provided Cryptographic Keys After 25 Years

Anthropic says an internal Claude research model reconstructed how the National Security Agency generated seeds…

1 day ago

Critical WordPress Flaws Allow Attackers to Execute Remote Code Without Authentication

Two WordPress Core vulnerabilities can be chained to execute remote code without authentication, according to…

1 day ago

OpenBSD Kernel Flaw Dating Back 37 Years Enables Local Privilege Escalation to Root

A newly disclosed OpenBSD kernel exploitation chain combines a shared memory accounting flaw with a…

1 day ago

GhostAction Hackers Hide Credential-Stealing Malware in Fake Font Files Across 94 GitHub Repositories

GhostAction attackers have expanded their GitHub credential theft campaign with a second stage that disguises…

1 day ago

SideCopy Hackers Target Indian Organizations With ReverseRAT Malware to Steal Data

Indian authorities have warned that SideCopy hackers are targeting Indian organizations, including academic institutions, with…

1 day ago