{"id":"CVE-2026-65120","kind":"vulnerability","aliases":["CVE-2026-65120"],"related":[],"title":"NVIDIA TensorRT-LLM: server-side request forgery lets attackers reach internal services","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.","summary":"NVIDIA TensorRT-LLM, an LLM inference library, has a server-side request forgery flaw. An attacker can make the server send requests to destinations of their choosing, which may leak information.","severity":"high","cvss":7.1,"epss":null,"exploited":false,"exploited_at":null,"products":[],"ecosystems":[],"tags":["ai-inference","ai-tooling","nvidia","self-hosted","tensorrt-llm"],"not_affected_if":["not-using-tensorrt-llm"],"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: the sources do not say which request path or setting is affected.","fix":"- Upgrade TensorRT-LLM to the patched version listed in NVIDIA's security bulletin, which the sources do not name.\n- unknown: the sources do not give a fixed version number.","status":"open","published_at":"2026-10-27T13:16:59+00:00","first_seen_at":"2026-10-27T13:31:51+00:00","triaged_at":"2026-10-27T13:32:18+00:00","updated_at":"2026-10-27T13:32:18+00:00","category":"ai-tooling","confidence":0.5,"triage":{"at":"2026-10-27T13:32:18Z","model":"claude-haiku-5-5","output":{"fix":"- Upgrade TensorRT-LLM to the patched version listed in NVIDIA's security bulletin, which the sources do not name.\n- unknown: the sources do not give a fixed version number.","reach":"unknown","title":"NVIDIA TensorRT-LLM: server-side request forgery lets attackers reach internal services","impact":"medium","summary":"NVIDIA TensorRT-LLM, an LLM inference library, has a server-side request forgery flaw. An attacker can make the server send requests to destinations of their choosing, which may leak information.","category":"ai-tooling","nicknames":[],"who_cares":["tensorrt-llm","nvidia","self-hosted","ai-inference"],"confidence":0.5,"needs_auth":"unknown","worst_case":"Server sends attacker-chosen requests, leaking internal information","analyst_note":"Find the NVIDIA security bulletin for TensorRT-LLM and confirm the affected versions, fixed version, and triggering endpoint.","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: the sources do not say which request path or setting is affected.","needs_analyst":true,"not_affected_if":["not-using-tensorrt-llm"],"who_is_affected":"Teams that run TensorRT-LLM to serve models, especially where the server accepts requests that make it fetch URLs or external resources. The sources do not say which endpoints or configurations trigger the flaw. Teams not running TensorRT-LLM can skip it.","user_interaction":"unknown","needs_special_setup":"unknown","severity_if_unknown":"high"},"cost_usd":0.000626,"input_tokens":3281,"output_tokens":596,"severity_before":"high"},"analyst_status":null,"analyst_note":null,"search":"'-2026':16A,19A '-65120':17A,20A 'ai':22B,25B 'ai-infer':21B 'ai-tool':24B 'attack':11A,51B,76C 'caus':78C 'choos':62B 'contain':71C 'could':77C 'cve':15A,18A 'destin':59B 'disclosur':95C 'exploit':87C 'flaw':49B 'forgeri':9A,48B,84C 'host':30B 'infer':23B,40B 'inform':66B,94C 'intern':13A 'lead':92C 'leak':65B 'let':10A 'librari':41B 'llm':4A,33B,37B,39B,70C 'make':53B 'may':64B 'might':91C 'nvidia':1A,27B,34B,67C 'reach':12A 'request':8A,47B,57B,83C 'self':29B 'self-host':28B 'send':56B 'server':6A,45B,55B,81C 'server-sid':5A,44B,80C 'servic':14A 'side':7A,46B,82C 'success':86C 'tensorrt':3A,32B,36B,69C 'tensorrt-llm':2A,31B,35B,68C 'tool':26B 'vulner':73C,90C","nicknames":[],"mitigations":[],"who_affected":"Teams that run TensorRT-LLM to serve models, especially where the server accepts requests that make it fetch URLs or external resources. The sources do not say which endpoints or configurations trigger the flaw. Teams not running TensorRT-LLM can skip it.","tier":"B","verified":{},"analyst_revisit_at":null,"first_public_at":"2026-10-27T13:16:59+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-2026-65120","url":"https://nvd.nist.gov/vuln/detail/CVE-2026-65120","severity":null,"published_at":"2026-10-27T13:16:59+00:00","modified_at":"2026-10-27T13:16:59+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"}