← back to feed
ETK-2026-1784Zero's internal tracking ID, independent of the CVE number.

ETK-2026-1784Zero's internal tracking ID, independent of the CVE number.

validatedZero produced an exploit or a check.

GHSA-jv65-rccw-j2j6 ↗

check availableZero validated a working check for it.proven exploitableZero has validated a full working exploit.
awaiting cvepublic pocExploit code has been published by someone. Source signals can also carry a PoC (Proof of Concept).kevAttacks confirmed in the real world.
pip:xinference >= 2.12.0, < 3.5.0
fixed in 3.5.0
CWE-94 Improper Control of Generation of Code ('Code Injection')
ETK-2026-1784Zero's internal tracking ID, independent of the CVE number., no cve assigned
pkg:pypi/xinference
cvss9.8cna vendorCVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
−4.5dTTV
25mAI analysisAnalysis: how long the AI spent reversing this signal.
Check if you are vulnerable
timeline
0 = trigger signal
trigger signal +0dcheck + exploit+0.5d
0
+1d
+2d
+3d
+4d
+5d

description

Xinference versions 2.12.0 through 3.4.0 improperly trust a client-controlled custom-model provenance value. An attacker who can access model registration and launch functions can cause code to execute with the model worker's operating-system permissions; this may require credentials in default 3.x deployments, but can be unauthenticated when the management API is exposed without authentication.

mitigation

Upgrade every Xinference supervisor and worker to version 3.5.0 or later. Before restarting or launching saved custom models, review persisted custom-model registrations for unexpected built-in markings; restrict management API access to trusted administrators until that review is complete.

fixed in 3.5.0

check
available
exploit
yes
automationDeterministic: a check that runs on its own, the same request on every target. Contextual: an agentic technique that needs to be applied per target, for bugs that depend on contextual flows.
deterministic
auth requiredWhether an attacker needs an account first, based on the CVSS vector's Privileges Required.
no
EPSSEstimate of the chance this CVE is exploited in the next 30 days. Refreshed daily.
–
agent run time
25m
cve.org record, t = 0The moment cve.org publishes the CVE in full.
not published yet
Δ to public pocExploit code has been published by someone. Source signals can also carry a PoC (Proof of Concept).
–
Δ to kevAttacks confirmed in the real world.
–

outcome

Differential: the exploit fired on the vulnerable build and stayed silent on the patched one.

source signals

Public signals that triggered or updated the pipeline. These are inputs for the deltas above.

  1. 2026-10-16 05:02Z

this entry in the feed

GET /api/v1/feed/ETK-2026-1784

{
  "etk_id": "ETK-2026-1784",
  "cve_id": null,
  "display_id": "ETK-2026-1784",
  "ghsa_id": "GHSA-jv65-rccw-j2j6",
  "euvd_id": null,
  "status": "published",
  "anchor_state": "awaiting_cve",
  "title": "Custom model provenance validation flaw in Xinference",
  "summary": "Xinference versions 2.12.0 through 3.4.0 improperly trust a client-controlled custom-model provenance value. An attacker who can access model registration and launch functions can cause code to execute with the model worker's operating-system permissions; this may require credentials in default 3.x deployments, but can be unauthenticated when the management API is exposed without authentication.",
  "mitigation": "Upgrade every Xinference supervisor and worker to version 3.5.0 or later. Before restarting or launching saved custom models, review persisted custom-model registrations for unexpected built-in markings; restrict management API access to trusted administrators until that review is complete.",
  "vendor": "pip",
  "product": "pip:xinference",
  "affected": [
    {
      "affected_notes": [],
      "affected_ranges": [
        ">= 2.12.0, < 3.5.0"
      ],
      "ecosystem": "pip",
      "fixed_versions": [
        "3.5.0"
      ],
      "package_name": "xinference",
      "purl": "pkg:pypi/xinference"
    }
  ],
  "cwes": [
    94
  ],
  "cvss_score": 9.8,
  "cvss_src": "cna_vendor",
  "cvss_vector": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H",
  "epss_score": null,
  "epss_percentile": null,
  "epss_asof": null,
  "automation": "deterministic",
  "check": true,
  "check_confidence_basis": "differential",
  "exploit": true,
  "pipeline_minutes": 24.750464316666665,
  "t0_at": null,
  "signals": {
    "check": {
      "at": "2026-10-16T15:56:03.578036+00:00",
      "days": -4.511058545081019,
      "is_lower_bound": true,
      "measured_at": "2026-10-21T04:11:59.323499+00:00",
      "url": null
    },
    "exploit": {
      "at": "2026-10-16T15:56:03.578036+00:00",
      "days": -4.511058545081019,
      "is_lower_bound": true,
      "measured_at": "2026-10-21T04:11:59.323499+00:00",
      "url": null
    },
    "trigger_signal": {
      "at": "2026-10-16T05:02:30+00:00",
      "days": -4.9649194019791665,
      "is_lower_bound": true,
      "measured_at": "2026-10-21T04:11:59.323499+00:00",
      "url": "https://github.com/xorbitsai/inference/security/advisories/GHSA-jv65-rccw-j2j6"
    }
  }
}

published 2026-10-16 15:31Z
last change 2026-10-21 03:59Z

Need the technical details behind an entry?

Every entry here shows what we attempted and when. The technical details, the generated exploit that fired it and the evidence behind the verdict are not public. Tell us about your use case and we’ll get back to you about extended API access.

Protected by reCAPTCHA; the Google Privacy Policy and Terms of Service apply.