Security assessments indicate that CVE-2026-54236 presents a notable risk, potentially requiring prompt mitigation.
Status: Received on 22 Jun 2026, 23:16 UTC
Published on: 22 Jun 2026, 23:16 UTC
CVSS Release: version 3
security-advisories@github.com
Secondary
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:N/A:N
CVE-2026-54236: vLLM is an inference and serving engine for large language models (LLMs). Prior to 0.23.1rc0, the fix for CVE-2026-22778, which introduced a sanitize_message helper that strips object-repr memory addresses from error messages before they reach the client, is incomplete: several response paths echo str(exc) directly to clients without calling sanitize_message. The unsanitized sites include the Anthropic API router in vllm/entrypoints/anthropic/api_router.py (the POST /v1/messages and POST /v1/messages/count_tokens handlers), the Server-Sent Events streaming converter in vllm/entrypoints/anthropic/serving.py, and the realtime speech-to-text WebSocket in vllm/entrypoints/speech_to_text/realtime/connection.py. These paths catch the exception inside the route coroutine and construct the JSONResponse themselves, bypassing the sanitizing global FastAPI exception handler, and WebSocket frames do not traverse that handler chain at all. Using the same primitive as the parent issue, an unauthenticated attacker can send malformed image bytes through the Anthropic Messages API image content parts so that PIL.Image.open raises an UnidentifiedImageError whose message contains the BytesIO object repr, leaking the heap memory address verbatim in the error.message field of the response body. This vulnerability is fixed in 0.23.1rc0.
The exploitability of CVE-2026-54236 depends on two key factors: attack complexity (the level of effort required to execute an exploit) and privileges required (the access level an attacker needs).
With low attack complexity and no required privileges, CVE-2026-54236 is an easy target for cybercriminals. Organizations should prioritize immediate mitigation measures to prevent unauthorized access and data breaches.
A lower complexity and fewer privilege requirements make exploitation easier. Security teams should evaluate these aspects to determine the urgency of mitigation strategies, such as patch management and access control policies.
Attack Complexity (AC) measures the difficulty in executing an exploit. A high AC means that specific conditions must be met, making an attack more challenging, while a low AC means the vulnerability can be exploited with minimal effort.
Privileges Required (PR) determine the level of system access necessary for an attack. Vulnerabilities requiring no privileges are more accessible to attackers, whereas high privilege requirements limit exploitation to authorized users with elevated access.
Above is the CVSS Sub-score Breakdown for CVE-2026-54236, illustrating how Base, Impact, and Exploitability factors combine to form the overall severity rating. A higher sub-score typically indicates a more severe or easier-to-exploit vulnerability.
Below is the Impact Analysis for CVE-2026-54236, showing how Confidentiality, Integrity, and Availability might be affected if the vulnerability is exploited. Higher values usually signal greater potential damage.
Stay updated with real-time CVE vulnerabilities and take action to secure your systems. Enhance your cybersecurity posture with the latest threat intelligence and mitigation techniques. Develop the skills necessary to defend against CVEs and secure critical infrastructures. Join the top cybersecurity professionals safeguarding today's infrastructures.