No attacks reportedCVE published · 8 Oct 2026
Ollama
CVE-2026-103663 Severity · Critical
CVSS 9.4
Affected
0.34.2 and later, below 0.35.0; older versions: status unknown
Fixed in
0.35.0 and later (latest release: 0.40.10)

CVE alert · AI and automation tools

CVE-2026-103663: Ollama trusts file names from model servers

An attacker can make Ollama write files outside its model folder and run code as root (full admin rights) on the default Docker image. Update to 0.35.0 or newer today. The CVE record does not declare versions before 0.34.2 safe.

Check your systems with your AI agent

Paste this prompt into an AI agent that can run commands on your machines. It only reads versions and settings, asks before using administrator rights, and ends with a report.

Check whether my systems are affected by CVE-2026-103663 in Ollama, then write me a report.

Read the alert first: https://securityaudit.ae/cve/cve-2026-103663-ollama/

What the alert says (facts checked 23 October 2026):
- Affected: 0.34.2 and later, below 0.35.0; older versions: status unknown
- Fixed in: 0.35.0 and later (latest release: 0.40.10)
- Exploited: No attacks reported as of 23 October 2026 (not in CISA's KEV catalog); public attack code since 8 October 2026; ready-to-run kit on GitHub since 19 October 2026

Rules:
- Read only. Do not install, update, restart, delete or change anything, and never run exploit or proof-of-concept code.
- Run only commands that read versions or settings. Ask me before any command that needs sudo or administrator rights.
- If you do not know which machines, containers, cloud accounts or sites are in scope, ask me for the list before you start.
- If a machine cannot be reached or a command fails, say so. Do not guess.

Steps:
1. Find where Ollama runs: servers, containers and their hosts, cloud services, sites.
2. On each one, read the installed version and the settings the alert names. The alert gives these commands:

    ollama -v

    curl http://localhost:11434/api/version

    docker exec -it ollama ollama -v

3. Compare each version with the affected and fixed versions above. When a vendor or a distribution ships its own fixed package, the alert names it.
4. Report: one line per machine or service with the version found, Affected, Not affected or Unknown, and the evidence (the command and its output). Then the actions to take, in order, from the alert's "How to fix" section, and what you could not check.

Key facts

FactDetail
ProductOllama (AI model server)
Affected versions0.34.2 and later, below 0.35.0; older versions: status unknown
Fixed in0.35.0 and later (latest release: 0.40.10)
Severity9.4 Critical (CVSS 4.0, CERT Polska)
Who can attackAnyone who can connect to port 11434 and run a model registry your server can reach; no login needed
ExploitedNo attacks reported as of 23 October 2026 (not in CISA's KEV catalog); public attack code since 8 October 2026; ready-to-run kit on GitHub since 19 October 2026
CVE published8 October 2026

What an attacker can do

Ollama runs AI models on your own servers. Its HTTP API listens on port 11434 and has no login.

The flaw is a path traversal (a file name with ../ that escapes its folder) in model downloads (POST /api/pull). Ollama reads a model's file list from a registry (a server that hosts models) and does not check the file names. So an attacker's registry can make your server write files outside its model folder.

On the default Docker image, the attacker plants a malicious library file in /usr/lib/ollama, which Ollama loads at the next restart. The attacker can then run any command as root inside the container (an isolated space on the server). CERT Polska, the Polish security team that registered this flaw, rates the impact on confidentiality, integrity and availability as High.

Who is affected

The CVE record lists Ollama 0.34.2 and later, below 0.35.0.

  • Official Docker image: Ollama runs as root and can write to /usr/lib/ollama. It listens on every network interface of the container (OLLAMA_HOST=0.0.0.0:11434). With the documented -p 11434:11434, Docker opens that port on all the server's networks. CERT Polska scores this case 9.4 Critical (CVSS 4.0, the standard severity scale out of 10).
  • Standard Linux install with install.sh: the service runs as the user ollama and does not reach root this way. The attacker can still create files outside the model folder, or wipe the content of existing ones. This case scores 6.9 Medium.

The attacker needs access to port 11434, and your server must reach a registry the attacker controls (the public attack code uses plain HTTP). A native install listens on 127.0.0.1 (the machine itself) by default. On some servers with a GPU, the planted code runs without a restart.

How to check your version

Run one of these commands. On the server that runs Ollama:

shell
ollama -v

Over the API, replace localhost with your server address:

shell
curl http://localhost:11434/api/version

The answer looks like {"version": "..."}.

On a Docker install, replace ollama with your container name:

shell
docker exec -it ollama ollama -v

Your server is vulnerable if the version is 0.34.2 or later and below 0.35.0. Update older versions too, because the CVE record gives them the status "unknown" and does not declare them safe.

Ollama published no advisory or workaround of its own. The notes of the patched release do not mention the flaw either, so update without waiting for a security bulletin.

What to do now

Update Ollama to 0.35.0 or later. On Linux, run the install script again:

shell
curl -fsSL https://ollama.com/install.sh | sh

On macOS and Windows, the app downloads updates: click "Restart to update". On Docker, download the image ollama/ollama:0.35.0 or newer and recreate the container.

Keep port 11434 off the internet and off untrusted networks. For remote access, put a proxy server that asks for a login in front of the API.

Neither Ollama nor Striga, the company that found the flaw, listed signs of compromise, but the company's write-up (dated 8 October 2026) and attack kit (on GitHub since 19 October 2026) give two leads:

  • an mlx_* folder or a libmlxc.* file in /usr/lib/ollama that the official image does not ship (the attack code plants mlx_zzz/libmlxc.tmp);
  • model pulls from unknown registries with "insecure":true, the plain HTTP option, in request bodies captured by a reverse proxy or a network capture. At the default log level, Ollama's own log shows only POST "/api/pull", without the model name or the insecure flag. With OLLAMA_DEBUG=1 set, look for a downloading blob line whose digest contains ../ or does not start with sha256.

Get help

Updating closes this flaw. Flaws in your own code get no CVE and no update, and a test of your app is how they are found.

AI agent and chatbot security audit

Can a message, an email or a document make your AI agent act against you? We try, and show you how to stop it.

From AED 35,000, excl. 5% VAT

Ask for a quote Book a call

Sources

  1. CVE-2026-103663 record CVE Program
  2. Vulnerability in Ollama software (CVE-2026-103663) CERT Polska
  3. DirtyDigest: A crafted manifest turns Ollama's model pull into an arbitrary file write and root RCE Striga (finder)
  4. striga-ai/CVE-2026-103663 reproduction kit Striga (finder)
  5. Ollama v0.35.0 release Ollama
  6. Known Exploited Vulnerabilities Catalog CISA

More CVE alerts

All CVE alerts