TRENDING
Dew drops strung along the threads of a spider web against a dark background, like nodes and edges in a graph
October 31, 2026
How to Detect Procurement Fraud Rings in Python With NetworkX and SQLite and Measure Each Rule’s Precision and Recall
Red hotel door hanger reading Do Not Disturb in four languages on a brass doorknob, a request that does not lock the door
October 31, 2026
OpenAI’s Shared Training Signal Turns robots.txt Into a Purpose Token From a Same-Day Individual Draft, and a Group That Blocks One Folder Reads as a Preference for Training on Every Other Path
A white extension cord with a Europlug at one end and a socket at the other, coiled and held together by a green strap
October 31, 2026
Cursor’s Marketplace Served a Tampered Claude Code Extension With an SSH Backdoor for 11 Hours, and Its Notice Now Names Two Versions Instead of One
Two open pages of an eighteenth-century ship's logbook with ruled columns and handwritten daily remarks
October 31, 2026
How to Build a Tamper-Evident Audit Log for AI Agent Actions in Python With Hash Chains, Signed Checkpoints, and Key Evolution
A large calico cat lying beside window blinds with its tongue out mid-hiss, a pun on Le Chonk and a model's refusals
October 31, 2026
Mistral’s Le Chonk Turns Cyber Defense Into a Download, but the Independent Index It Cites Ranks It Seventh, and Its One First Place Among Open Models Is Won by a Single Task
31 Oct 2026
SXZ.io SXZ.io
  • Home
Search the Site
Popular Searches:
Technology Amazon AI
Recent Posts
Carved wooden cuckoo bird perched above the leaf-carved roof and small door of a Black Forest cuckoo clock
How to Build a Cuckoo Filter in Python So a Fast Membership Check Can Forget Deleted Items
October 31, 2026
Printed page of text with a white strip of correction fluid and a handwritten correction in blue ink
OpenAI’s Model Spec Rewrite Turns Style Into a Preference and Oversight Into a Root Rule, but the Public Evals Still Test the December Text
October 31, 2026
A gravel track on open moorland splits into several branches beside a wooden signpost
Anthropic Says Claude Wrote a Remote Root Exploit for a Linux Kernel Bug That Two One-Line Changes Fix, and Red Hat and Ubuntu Have Yet to Ship the Patched Kernel
October 31, 2026
SXZ.io SXZ.io
  • Home

Categories

Articles 275 Posts
News 278 Posts
Learning Hub 248 Posts
Home/News/Attackers Exploit a Maximum-Severity GitLab Flaw as CISA’s Federal Deadline Lands Today
News

Attackers Exploit a Maximum-Severity GitLab Flaw as CISA’s Federal Deadline Lands Today

A maximum-severity GitLab flaw is already being exploited, and it is the second critical GitLab bug targeted within days of disclosure in a matter of weeks.

September 14, 2026 3 Min Read
94

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has given federal agencies until today, September 14, to patch a maximum-severity vulnerability in GitLab’s self-managed software. The flaw, tracked as CVE-2026-85706, carries the highest possible CVSS score, 10.0 out of 10, and lets an attacker with no credentials at all read arbitrary files off a vulnerable GitLab server. Attackers started probing for it within a day of GitLab shipping a fix, the second time in a matter of weeks that a critical GitLab bug has been targeted almost as soon as it became public.

Table Of Content

  • An Unauthenticated Path to Arbitrary Files
  • From Patch to Probing in a Day
  • A Second Critical GitLab Bug in Weeks
  • What to Do Now

An Unauthenticated Path to Arbitrary Files

GitLab disclosed the flaw in its own advisory for its September 10 Critical Patch Release, describing it as a path traversal issue in the repository commits API: “GitLab has remediated an issue that, under certain conditions, an unauthenticated user could have read arbitrary files from the GitLab server due to improper path confinement and missing authentication enforcement in the repository commits API.” The vulnerability’s CVSS 3.1 vector, AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:N, explains why it maxed out the scale: the attack is reachable over the network, needs no special skill or privileges, requires no user interaction, and its confidentiality and integrity impact both rate high.

All self-managed GitLab Community Edition and Enterprise Edition versions from 18.7 up to (but not including) 19.1.8, 19.2.6, and 19.3.2 are affected. GitLab.com and GitLab Dedicated customers were already running the patched version and do not need to take any action. GitLab credited researcher s3ntago with reporting the bug through its HackerOne bug bounty program. Because the affected endpoint requires no authentication, threat intelligence firm watchTowr warned that a single request against the repository commits API could return source code, CI/CD secrets, credentials, or other files the server was never meant to expose.

From Patch to Probing in a Day

GitLab shipped versions 19.3.2, 19.2.6, and 19.1.8 on September 10. By the next day, watchTowr’s threat intelligence unit said it was already picking up “behavioral probes for this vulnerability against Attacker Eye, our global honeypot network, meaning attackers have already successfully reverse engineered and reproduced the vulnerability.” The firm assessed with high confidence that the bug would “rapidly transition to indiscriminate, in-the-wild exploitation given the low complexity of exploitation.”

CISA added CVE-2026-85706 to its Known Exploited Vulnerabilities (KEV) catalog on September 11, one day after GitLab’s patch, and flagged it as requiring forensic triage, BleepingComputer first reported. Under Binding Operational Directive (BOD) 26-04, which sets federal patching deadlines by risk, that gave civilian federal agencies a three-day window to remediate, due today. BOD 26-04 legally binds only Federal Civilian Executive Branch agencies, but CISA said in its advisory that “these types of vulnerabilities are frequent attack vectors for malicious cyber actors and pose significant risks to the federal enterprise,” and it encouraged every organization running self-managed GitLab to treat the KEV listing as a signal to patch now, not just federal ones.

A Second Critical GitLab Bug in Weeks

In August, sxz.io covered GitLab’s emergency patch for CVE-2026-19478, an unauthenticated GraphQL code injection bug that scored 9.4 on the same CVSS scale and also needed an out-of-band fix. Jake Knott, head of threat intelligence at watchTowr, drew the connection directly in remarks to SecurityWeek: “This is the second instance of a critical-severity GitLab vulnerability in recent weeks, following the previous GraphQL code injection (CVE-2026-19478) that was almost immediately actively exploited.” He added that the appeal to attackers is obvious, since unauthorized access to a GitLab instance can expose source code, CI/CD secrets, and credentials, and can let an attacker inject code into build pipelines to poison anything downstream of them.

GitLab’s September 10 release was a large one. Alongside CVE-2026-85706, it fixed 17 other security issues, including a second critical bug, CVE-2026-87719 (CVSS 9.9), an insecure deserialization flaw in the GraphQL subscription serializer that could let an authenticated user with Duo Chat access obtain Advanced Search instance configurations and sensitive credentials through a crafted GraphQL subscription argument. Six more issues in the same release rated high severity, covering everything from remote code execution during project imports to bypasses of protected CI/CD variables.

What to Do Now

GitLab and watchTowr both recommend upgrading every self-managed GitLab CE or EE instance to 19.1.8, 19.2.6, or 19.3.2 (or later) immediately, and removing public internet access to any instance that cannot be patched right away. Defenders who suspect they may already be compromised should hunt through access logs for HTTP POST requests to /api/v4/projects/{id}/repository/commits/ URIs that include a file.path parameter, the specific pattern both GitLab and watchTowr flagged as the exploitation signature for this bug.

Tags:

CISADevOps SecurityGitLabSoftware Supply Chain SecurityVulnerability Management

Share

An open 18th-century accounting ledger book showing handwritten rows and columns of numbers in ink
Previous Post

How to Parse CSV Files in Python and Survive Quotes, Encodings, and Bad Rows

A woven basket filled with a mix of brown, tan, white, and speckled quail eggs sitting in straw
Next Post

Insight Partners’ Deven Parekh Turns AI Concentration Into a Diversification Bet

No Comment! Be the first one.

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Latest
31 Oct
How to Build a Cuckoo Filter in Python So a Fast Membership Check Can Forget Deleted Items
31 Oct
OpenAI’s Model Spec Rewrite Turns Style Into a Preference and Oversight Into a Root Rule, but the Public Evals Still Test the December Text
Trending
October 31, 2026
How to Build a Cuckoo Filter in Python So a Fast Membership Check Can Forget Deleted Items
October 31, 2026
OpenAI’s Model Spec Rewrite Turns Style Into a Preference and Oversight Into a Root Rule, but the Public Evals Still Test the December Text
October 31, 2026
Anthropic Says Claude Wrote a Remote Root Exploit for a Linux Kernel Bug That Two One-Line Changes Fix, and Red Hat and Ubuntu Have Yet to Ship the Patched Kernel
October 30, 2026
How to Detect Procurement Fraud Rings in Python With NetworkX and SQLite and Measure Each Rule’s Precision and Recall
October 30, 2026
OpenAI’s Shared Training Signal Turns robots.txt Into a Purpose Token From a Same-Day Individual Draft, and a Group That Blocks One Folder Reads as a Preference for Training on Every Other Path
October 30, 2026
Cursor’s Marketplace Served a Tampered Claude Code Extension With an SSH Backdoor for 11 Hours, and Its Notice Now Names Two Versions Instead of One

Related Posts

Rows of server racks in a data center representing network infrastructure targeted by botnets
News

C0XMO Botnet Shows Why Old Router Firmware Still Matters

June 7, 2026
Close-up of a USB flash drive, representing physical data-theft risk in office security incidents
News

Fake IT Support Is Now Walking Through the Front Door

June 7, 2026
A phone security app on a smartphone resting on a laptop keyboard.
News

Everest Forms Pro Flaw Is Being Exploited to Create Rogue WordPress Admins

June 7, 2026
A phone secured by a padlock, illustrating AI data-leak containment and security controls.
News

OpenAI’s Lockdown Mode Is a Data-Leak Brake, Not a Prompt-Injection Cure

June 8, 2026
SXZ.io SXZ.io
  • [email protected]

Categories

Articles
Learning Hub
News

All Rights Reserved by SXZ.io ©2026