<?xml version='1.0' encoding='UTF-8'?>
<?xml-stylesheet href="/static/style.xsl" type="text/xsl"?>
<rss xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/" version="2.0">
  <channel>
    <title>Most recent entries from all</title>
    <link>https://cve.radiocsirt.org</link>
    <description>Contains only the most 10 recent entries.</description>
    <docs>http://www.rssboard.org/rss-specification</docs>
    <generator>python-feedgen</generator>
    <language>en</language>
    <lastBuildDate>Mon, 16 Nov 2026 21:32:18 +0000</lastBuildDate>
    <item>
      <title>EUVD-2026-396349</title>
      <link>https://cve.radiocsirt.org/vuln/euvd-2026-396349</link>
      <description>EUVD-2026-396349</description>
      <guid isPermaLink="false">https://cve.radiocsirt.org/vuln/euvd-2026-396349</guid>
    </item>
    <item>
      <title>fkie_cve-2026-115914</title>
      <link>https://cve.radiocsirt.org/vuln/fkie_cve-2026-115914</link>
      <description>&lt;p&gt;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.&lt;/p&gt;</description>
      <content:encoded>&lt;p&gt;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.&lt;/p&gt;</content:encoded>
      <guid isPermaLink="false">https://cve.radiocsirt.org/vuln/fkie_cve-2026-115914</guid>
    </item>
    <item>
      <title>GHSA-gq5q-89c7-vhvx — vLLM DeepSeek-OCR image processing vulnerable to denial of service</title>
      <link>https://cve.radiocsirt.org/vuln/ghsa-gq5q-89c7-vhvx</link>
      <description>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: vllm&lt;/p&gt;
&lt;p&gt;## Summary&lt;/p&gt;
&lt;p&gt;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.&lt;/p&gt;
&lt;p&gt;The image model&amp;#39;s input validation contains a guard inherited from the upstream DeepSeek-OCR reference implementation:&lt;/p&gt;
&lt;p&gt;```python
# vllm/model_executor/models/deepseek_ocr.py:478-479  (_parse_and_validate_image_input)
if torch.sum(pixel_values).item() == 0:
    return None
```&lt;/p&gt;
&lt;p&gt;In the original research code this was an empty-batch shortcut; in vLLM&amp;#39;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&amp;#39;s 2²⁴ exactness bound). `ImageOps.pad(image, (1024, 1024))` is the identity map for an already-1024×1024 image (`pr…&lt;/p&gt;</description>
      <content:encoded>&lt;p&gt;&lt;strong&gt;Affected:&lt;/strong&gt; PyPI: vllm&lt;/p&gt;
&lt;p&gt;## Summary&lt;/p&gt;
&lt;p&gt;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.&lt;/p&gt;
&lt;p&gt;The image model&amp;#39;s input validation contains a guard inherited from the upstream DeepSeek-OCR reference implementation:&lt;/p&gt;
&lt;p&gt;```python
# vllm/model_executor/models/deepseek_ocr.py:478-479  (_parse_and_validate_image_input)
if torch.sum(pixel_values).item() == 0:
    return None
```&lt;/p&gt;
&lt;p&gt;In the original research code this was an empty-batch shortcut; in vLLM&amp;#39;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&amp;#39;s 2²⁴ exactness bound). `ImageOps.pad(image, (1024, 1024))` is the identity map for an already-1024×1024 image (`pr…&lt;/p&gt;</content:encoded>
      <guid isPermaLink="false">https://cve.radiocsirt.org/vuln/ghsa-gq5q-89c7-vhvx</guid>
    </item>
    <item>
      <title>WID-SEC-W-2026-4051 — vllm: Mehrere Schwachstellen</title>
      <link>https://cve.radiocsirt.org/vuln/wid-sec-w-2026-4051</link>
      <description>&lt;p&gt;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.&lt;/p&gt;</description>
      <content:encoded>&lt;p&gt;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.&lt;/p&gt;</content:encoded>
      <guid isPermaLink="false">https://cve.radiocsirt.org/vuln/wid-sec-w-2026-4051</guid>
    </item>
  </channel>
</rss>
