{"id":"CVE-2026-24269","kind":"vulnerability","aliases":["CVE-2026-24269"],"related":[],"title":"NVIDIA TensorRT-LLM: server-side request forgery leads to data disclosure and tampering","description":"NVIDIA TensorRT-LLM contains a vulnerability where an attacker could cause a server-side request forgery. A successful exploit of this vulnerability might lead to information disclosure and data tampering.","summary":"NVIDIA TensorRT-LLM, a library for serving large language models, has a server-side request forgery flaw. An attacker who can send crafted requests can make the server fetch internal resources, which may expose information and allow data tampering.","severity":"high","cvss":7.1,"epss":null,"exploited":false,"exploited_at":null,"products":[],"ecosystems":[],"tags":["ai-tooling","llm-serving","nvidia","self-hosted","tensorrt-llm"],"not_affected_if":["no-tensorrt-llm-deployment"],"how_to_check":"- Check the installed TensorRT-LLM version with `pip show tensorrt_llm` and compare it to the fixed release in NVIDIA's bulletin\n- unknown whether the sources name the exact affected version range","fix":"- Upgrade `tensorrt_llm` to the patched version listed in NVIDIA's security bulletin (version not given in the sources)\n- Restrict outbound requests from the inference server to required hosts until patched","status":"open","published_at":"2026-10-27T13:16:57+00:00","first_seen_at":"2026-10-27T13:31:51+00:00","triaged_at":"2026-10-27T13:32:20+00:00","updated_at":"2026-10-27T13:32:20+00:00","category":"ai-tooling","confidence":0.5,"triage":{"at":"2026-10-27T13:32:20Z","model":"claude-haiku-5-5","output":{"fix":"- Upgrade `tensorrt_llm` to the patched version listed in NVIDIA's security bulletin (version not given in the sources)\n- Restrict outbound requests from the inference server to required hosts until patched","reach":"network","title":"NVIDIA TensorRT-LLM: server-side request forgery leads to data disclosure and tampering","impact":"medium","summary":"NVIDIA TensorRT-LLM, a library for serving large language models, has a server-side request forgery flaw. An attacker who can send crafted requests can make the server fetch internal resources, which may expose information and allow data tampering.","category":"ai-tooling","nicknames":[],"who_cares":["tensorrt-llm","nvidia","self-hosted","llm-serving"],"confidence":0.5,"needs_auth":"unknown","worst_case":"Internal resources reachable from the server may be read and data altered","analyst_note":"Find NVIDIA's security bulletin for CVE-2026-24269 to get affected versions, the fixed release and the vulnerable endpoint or feature.","how_to_check":"- Check the installed TensorRT-LLM version with `pip show tensorrt_llm` and compare it to the fixed release in NVIDIA's bulletin\n- unknown whether the sources name the exact affected version range","needs_analyst":true,"not_affected_if":["no-tensorrt-llm-deployment"],"who_is_affected":"Teams running NVIDIA TensorRT-LLM to serve models, especially where the server accepts requests that cause it to fetch URLs or resources. Exposure depends on whether the request path reaches the flawed code; the sources do not say which features are affected.","user_interaction":"unknown","needs_special_setup":"unknown","severity_if_unknown":"high"},"cost_usd":0.000633,"input_tokens":3286,"output_tokens":609,"severity_before":"high"},"analyst_status":null,"analyst_note":null,"search":"'-2026':17A,20A '-24269':18A,21A 'ai':23B 'ai-tool':22B 'allow':73B 'attack':55B,85C 'caus':87C 'contain':80C 'could':86C 'craft':59B 'cve':16A,19A 'data':12A,74B,106C 'disclosur':13A,104C 'exploit':96C 'expos':70B 'fetch':65B 'flaw':53B 'forgeri':9A,52B,93C 'host':31B 'inform':71B,103C 'intern':66B 'languag':44B 'larg':43B 'lead':10A,101C 'librari':40B 'llm':4A,26B,34B,38B,79C 'llm-serv':25B 'make':62B 'may':69B 'might':100C 'model':45B 'nvidia':1A,28B,35B,76C 'request':8A,51B,60B,92C 'resourc':67B 'self':30B 'self-host':29B 'send':58B 'serv':27B,42B 'server':6A,49B,64B,90C 'server-sid':5A,48B,89C 'side':7A,50B,91C 'success':95C 'tamper':15A,75B,107C 'tensorrt':3A,33B,37B,78C 'tensorrt-llm':2A,32B,36B,77C 'tool':24B 'vulner':82C,99C","nicknames":[],"mitigations":[],"who_affected":"Teams running NVIDIA TensorRT-LLM to serve models, especially where the server accepts requests that cause it to fetch URLs or resources. Exposure depends on whether the request path reaches the flawed code; the sources do not say which features are affected.","tier":"B","verified":{},"analyst_revisit_at":null,"first_public_at":"2026-10-27T13:16:57+00:00","kev":null,"cvss_vector":"CVSS:3.1/AV:L/AC:L/PR:N/UI:N/S:C/C:H/I:N/A:N","cvss_by":"psirt@nvidia.com","poc_count":0,"sources":[{"source":"nvd","external_id":"CVE-2026-24269","url":"https://nvd.nist.gov/vuln/detail/CVE-2026-24269","severity":null,"published_at":"2026-10-27T13:16:57+00:00","modified_at":"2026-10-27T13:16:57+00:00","first_seen_at":"2026-10-27T13:31:51+00:00","fetched_at":"2026-10-27T13:31:51+00:00","note":null}],"documents":[],"notice":"Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai"}