<?xml version='1.0' encoding='UTF-8'?>
<?xml-stylesheet href="/static/style.xsl" type="text/xsl"?>
<feed xmlns="http://www.w3.org/2005/Atom" xml:lang="en">
  <id>https://cve.radiocsirt.org/rss/recent/all/10</id>
  <title>Most recent entries from all</title>
  <updated>2026-11-16T21:43:58.897470+00:00</updated>
  <author>
    <name>Vulnerability-Lookup</name>
    <email>csirt@opendfir.org</email>
  </author>
  <link href="https://cve.radiocsirt.org" rel="alternate"/>
  <generator uri="https://lkiesow.github.io/python-feedgen" version="1.0.0">python-feedgen</generator>
  <subtitle>Contains only the most 10 recent entries.</subtitle>
  <entry>
    <id>https://cve.radiocsirt.org/vuln/euvd-2026-396349</id>
    <title>EUVD-2026-396349</title>
    <updated>2026-11-16T21:43:58.900542+00:00</updated>
    <content>EUVD-2026-396349</content>
    <link href="https://cve.radiocsirt.org/vuln/euvd-2026-396349"/>
  </entry>
  <entry>
    <id>https://cve.radiocsirt.org/vuln/fkie_cve-2026-115914</id>
    <title>fkie_cve-2026-115914</title>
    <updated>2026-11-16T21:43:58.900572+00:00</updated>
    <content type="xhtml">
      <div xmlns="http://www.w3.org/1999/xhtml">
        <p>vLLM is an inference and serving engine for large language models. From 0.11.1 until 0.32.0, the DeepSeek-OCR, DeepSeek-OCR-2, and Unlimited-OCR image path allows a remote client that can reach the OpenAI-compatible API to submit a legal image whose normalized pixels sum to zero when that image is alone in its per-step encoder batch. _parse_and_validate_image_input returns no embedding, embed_multimodal propagates that value, and sanity_check_mm_encoder_outputs triggers a fatal assertion that shuts down the engine for all users. This issue is fixed in version 0.32.0.</p>
      </div>
    </content>
    <link href="https://cve.radiocsirt.org/vuln/fkie_cve-2026-115914"/>
  </entry>
  <entry>
    <id>https://cve.radiocsirt.org/vuln/ghsa-gq5q-89c7-vhvx</id>
    <title>GHSA-gq5q-89c7-vhvx — vLLM DeepSeek-OCR image processing vulnerable to denial of service</title>
    <updated>2026-11-16T21:43:58.900603+00:00</updated>
    <content type="xhtml">
      <div xmlns="http://www.w3.org/1999/xhtml"><p><strong>Affected:</strong> PyPI: vllm</p>
<p>## Summary</p>
<p>Any vLLM deployment serving a model of the DeepSeek-OCR family — `deepseek-ai/DeepSeek-OCR`, `deepseek-ai/DeepSeek-OCR-2`, or `baidu/Unlimited-OCR` (whose `UnlimitedOCRForCausalLM` subclasses `DeepseekOCRForCausalLM`) — can be **killed by a single unauthenticated chat request carrying an ordinary, schema-valid PNG image**, under completely default server configuration.</p>
<p>The image model's input validation contains a guard inherited from the upstream DeepSeek-OCR reference implementation:</p>
<p>```python
# vllm/model_executor/models/deepseek_ocr.py:478-479  (_parse_and_validate_image_input)
if torch.sum(pixel_values).item() == 0:
    return None
```</p>
<p>In the original research code this was an empty-batch shortcut; in vLLM's serving architecture `pixel_values` is the actual client image tensor, and the condition is **exactly satisfiable by a legitimate image**: the in-tree `DeepseekOCRProcessor` normalizes pixels with `ToTensor + Normalize(0.5, 0.5)` (`vllm/transformers_utils/processors/deepseek_ocr.py:120-140`), mapping pixel value 0 → exactly −1.0 and 255 → exactly +1.0 in float32. A 1024×1024 image that is half pure black and half pure white (a perfectly ordinary 1-bit-style scanned-document PNG for an OCR model) therefore yields `torch.sum(pixel_values) == 0.0` **exactly** (equal counts of ±1.0; pairwise reduction stays integer-exact, well inside float32's 2²⁴ exactness bound). `ImageOps.pad(image, (1024, 1024))` is the identity map for an already-1024×1024 image (`pr…</p></div>
    </content>
    <link href="https://cve.radiocsirt.org/vuln/ghsa-gq5q-89c7-vhvx"/>
  </entry>
  <entry>
    <id>https://cve.radiocsirt.org/vuln/wid-sec-w-2026-4051</id>
    <title>WID-SEC-W-2026-4051 — vllm: Mehrere Schwachstellen</title>
    <updated>2026-11-16T21:43:58.900685+00:00</updated>
    <content type="xhtml">
      <div xmlns="http://www.w3.org/1999/xhtml">
        <p>Ein Angreifer kann mehrere Schwachstellen in vllm ausnutzen, um beliebigen Programmcode auszuführen, um einen Denial of Service Angriff durchzuführen, um Informationen offenzulegen und um Sicherheitsvorkehrungen zu umgehen.</p>
      </div>
    </content>
    <link href="https://cve.radiocsirt.org/vuln/wid-sec-w-2026-4051"/>
  </entry>
</feed>
