{"items":[{"id":"NVI-00014175","kind":"vulnerability","title":"lmdeploy: Hardcoded trust_remote_code=True is an implicit unsafe remote-code load path with no user opt-out","revision":2,"unknowns":["Customer inventory, deployment, reachability and exploitation have not been inspected.","NVD lastModified does not establish freshness of every associated CPE record."],"lifecycle":"active","identifiers":["GHSA-9xq9-36w5-q796","CVE-2026-46517"],"plain_summary":"GitHub lists this reviewed advisory for lmdeploy. Published version ranges and source prerequisites are shown below; this is not an inspection of a visitor’s application.","review_status":"source_listed","control_topics":[],"public_actions":[{"summary":"Identify the exact component and deployed version, examine the linked publisher’s conditions, and retain evidence of any mitigation.","source_ids":["sr-0578340ce2167c061bec33a8ec671d29"],"action_type":"investigate","limitations":"Confirm current vendor support, compatibility and deployment evidence. This report has not tested a fix in your system.","current_supported_target":null}],"schema_version":"1.1.0","source_metrics":[{"kind":"CVSS","as_of":"2026-08-31T14:33:47.000Z","value":"7.8 high","meaning":"GitHub source severity; exploitability and local consequence require separate evidence.","source_id":"sr-551ed2c7f0c8c4f7097264abbe7f3a62"},{"kind":"CVSS","as_of":"2026-08-31T15:17:17.260Z","value":"7.8 HIGH (3.1)","meaning":"security-advisories@github.com severity; no combined Null risk score.","source_id":"sr-0578340ce2167c061bec33a8ec671d29"}],"why_it_matters":"Published affected context helps an owner decide what evidence to gather. This report has not examined your application.","related_sources":[{"id":"sr-551ed2c7f0c8c4f7097264abbe7f3a62","url":"https://github.com/advisories/GHSA-9xq9-36w5-q796","basis":"vendor_advisory","notes":"GitHub Advisory Database; CC-BY-4.0. Each advisory links its original attribution. Null normalizes fields.","title":"GHSA-9xq9-36w5-q796","publisher":"GitHub Advisory Database","observed_at":"2026-11-02T02:21:58.521Z"},{"id":"sr-0578340ce2167c061bec33a8ec671d29","url":"https://nvd.nist.gov/vuln/detail/CVE-2026-46517","basis":"official_catalog","notes":"This product uses data from the NVD API but is not endorsed or certified by the NVD.","title":"CVE-2026-46517","publisher":"NIST National Vulnerability Database","observed_at":"2026-11-08T01:47:36.551Z"}],"affected_context":[{"certainty":"source_listed","source_ids":["sr-551ed2c7f0c8c4f7097264abbe7f3a62"],"technology":"lmdeploy","known_exclusions":[],"conditions_summary":"> ## 📋 Reframing (2026-05-02): implicit unsafe remote-code path, not \"supply-chain\"\n>\n> The accurate description of this vulnerability is:\n> **\"`get_model_arch` and related helpers hardcode `trust_remote_code=True`\n> with no opt-out, creating an implicit unsafe remote-code load path\n> on every model fetch.\"**\n>\n> What this report does NOT claim:\n> * It is NOT a network-attack RCE — the user supplies the model\n>   reference; LMDeploy honors it.\n> * It is NOT a \"supply chain\" CVE in the classical sense (where a\n>   benign upstream is compromised) — the user explicitly types the\n>   repo name.\n>\n> What this report DOES claim:\n> * Other inference frameworks (vLLM, TGI, Hugging Face transformers\n>   itself) all expose `--trust-remote-code` as **opt-in** so that\n>   users who consciously load known-safe repos can opt in, while\n>   users following a tutorial cannot accidentally execute attacker\n>   Python by typing a wrong repo name.\n> * LMDeploy's hardcoded True is an **implicit** trust-boundary\n>   override that violates HF Transformers' default-secure stance\n>   (`trust_remote_code=False` since transformers ≥ 4.30).\n> * The fix is a one-line CLI flag (`--trust-remote-code`) defaulting","public_affected_ranges":["<= 0.12.3"]}],"null_published_at":"2026-11-08T19:53:44.017Z","source_updated_at":"2026-08-31T14:33:47.000Z","source_disclosed_at":"2026-05-21T19:33:32.000Z","is_synthetic_fixture":false,"null_last_reviewed_at":null,"public_change_summary":"Source facts changed; compare the preserved prior revision.","null_first_observed_at":"2026-11-02T02:21:58.521Z"},{"id":"NVI-00014175","kind":"vulnerability","title":"lmdeploy: Hardcoded trust_remote_code=True is an implicit unsafe remote-code load path with no user opt-out","revision":1,"unknowns":["Customer inventory, deployment, reachability and exploitation have not been inspected."],"lifecycle":"active","identifiers":["GHSA-9xq9-36w5-q796","CVE-2026-46517"],"plain_summary":"GitHub lists this reviewed advisory for lmdeploy. Published version ranges and source prerequisites are shown below; this is not an inspection of a visitor’s application.","review_status":"source_listed","control_topics":[],"public_actions":[{"summary":"Identify the exact component and deployed version, examine the linked publisher’s conditions, and retain evidence of any mitigation.","source_ids":["sr-551ed2c7f0c8c4f7097264abbe7f3a62"],"action_type":"investigate","limitations":"Confirm current vendor support, compatibility and deployment evidence. This report has not tested a fix in your system.","current_supported_target":null}],"schema_version":"1.1.0","source_metrics":[{"kind":"CVSS","as_of":"2026-08-31T14:33:47.000Z","value":"7.8 high","meaning":"GitHub source severity; exploitability and local consequence require separate evidence.","source_id":"sr-551ed2c7f0c8c4f7097264abbe7f3a62"}],"why_it_matters":"Published affected context helps an owner decide what evidence to gather. This report has not examined your application.","related_sources":[{"id":"sr-551ed2c7f0c8c4f7097264abbe7f3a62","url":"https://github.com/advisories/GHSA-9xq9-36w5-q796","basis":"vendor_advisory","notes":"GitHub Advisory Database; CC-BY-4.0. Each advisory links its original attribution. Null normalizes fields.","title":"GHSA-9xq9-36w5-q796","publisher":"GitHub Advisory Database","observed_at":"2026-11-02T02:21:58.521Z"}],"affected_context":[{"certainty":"source_listed","source_ids":["sr-551ed2c7f0c8c4f7097264abbe7f3a62"],"technology":"lmdeploy","known_exclusions":[],"conditions_summary":"> ## 📋 Reframing (2026-05-02): implicit unsafe remote-code path, not \"supply-chain\"\n>\n> The accurate description of this vulnerability is:\n> **\"`get_model_arch` and related helpers hardcode `trust_remote_code=True`\n> with no opt-out, creating an implicit unsafe remote-code load path\n> on every model fetch.\"**\n>\n> What this report does NOT claim:\n> * It is NOT a network-attack RCE — the user supplies the model\n>   reference; LMDeploy honors it.\n> * It is NOT a \"supply chain\" CVE in the classical sense (where a\n>   benign upstream is compromised) — the user explicitly types the\n>   repo name.\n>\n> What this report DOES claim:\n> * Other inference frameworks (vLLM, TGI, Hugging Face transformers\n>   itself) all expose `--trust-remote-code` as **opt-in** so that\n>   users who consciously load known-safe repos can opt in, while\n>   users following a tutorial cannot accidentally execute attacker\n>   Python by typing a wrong repo name.\n> * LMDeploy's hardcoded True is an **implicit** trust-boundary\n>   override that violates HF Transformers' default-secure stance\n>   (`trust_remote_code=False` since transformers ≥ 4.30).\n> * The fix is a one-line CLI flag (`--trust-remote-code`) defaulting","public_affected_ranges":["<= 0.12.3"]}],"null_published_at":"2026-11-02T10:00:37.218Z","source_updated_at":"2026-08-31T14:33:47.000Z","source_disclosed_at":"2026-05-21T19:33:32.000Z","is_synthetic_fixture":false,"null_last_reviewed_at":null,"public_change_summary":"First indexed by Null; source disclosure date is shown separately.","null_first_observed_at":"2026-11-02T02:21:58.521Z"}],"next_cursor":null,"snapshot_revision":"42594"}