CVE-2026-48797
Backpropagate: backprop ui --auth and backprop ui --share do not enforce authentication
- Published
- Updated
- Reserved
- Assigned by
- GitHub_M
Description
Backpropagate is a Python library for fine-tuning large language models on a single GPU. In versions 1.1.0 and 1.1.1, the optional Reflex web UI exposes a training control plane without authentication: dataset upload, model load, training start/stop, multi-run orchestration, GGUF export, and HuggingFace Hub push. The CLI accepts two operator-facing flags intended as security controls: --auth user:pass — documented as "require HTTP Basic authentication on every request to the UI." and--share — documented as "expose the UI on a public address; requires --auth." When --auth user:pass is passed, the CLI prints Auth: enabled (user: <username>) to confirm to the operator that authentication is active, then exports BACKPROPAGATE_UI_AUTH=user:pass to the subprocess that launches the Reflex backend. The Reflex backend (backpropagate/ui_app/**) never reads BACKPROPAGATE_UI_AUTH. No authentication middleware is registered. No request-level guard runs. No WebSocket upgrade guard runs. Any client that reaches the bound port — local or remote, depending on whether --share is used — has full UI access. An inline comment at backpropagate/cli.py:1217-1218 in the v1.1.0 source documents the gap: "For Phase 1 the variable is exported but Reflex doesn't read it yet." This comment was internal-facing; the user-facing documentation (README, CHANGELOG, SHIP_GATE) advertised the contract as enforced. An attacker who reaches the bound port can read uploaded datasets, trigger arbitrary training runs against any local base models as well as read their paths, trigger HuggingFace Hub pushes and cause disk-fill DoS. This issue has been fixed in version 1.2.0. If developers cannot immediately upgrade to 1.2.0 run backprop ui with no flags so it binds to localhost, use SSH port-forwarding (ssh -L 7860:localhost:7860 <training-host>) instead of --share for remote access, and audit any host previously launched with --share, re-issuing any HF tokens used during those sessions.
Scores
| Scored by | CVSS | Score | Vector |
|---|---|---|---|
| GitHub_MCNA (assigner) | 4.0 | Critical9.3 | CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N |
| NVDNVD · Deferred | 4.0 | Critical9.3 | CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X |
- EPSS (exploitation in 30 days)
- 0.32% · top 74% of all CVEs (6 Nov 2026)
- CISA SSVC
- Exploitation: none · Automatable: no · Technical impact: total
The list uses the assigner's score when there is one, then CISA's, then NVD's: Critical 9.3.
Weakness
- CWE-358: Improperly Implemented Security Check for Standard (opens in a new tab)All about CWE-358 and its CVEs
- CWE-862: Missing Authorization (opens in a new tab)All about CWE-862 and its CVEs
- CWE-1295: Debug Messages Revealing Unnecessary Information (opens in a new tab)All about CWE-1295 and its CVEs
Affected products
mcp-tool-shop-org · backpropagate
- affected >= 1.1.0, < 1.2.0
mcp-tool-shop-org · @mcptoolshop/backpropagate
- affected >= 1.1.0, < 1.2.0
Affected packages
From reviewed advisories of the GitHub Advisory Database.
- PyPIbackpropagateGHSA-f65r-h4g3-3h9h (opens in a new tab)
- npm@mcptoolshop/backpropagateGHSA-f65r-h4g3-3h9h (opens in a new tab)
References (2)
Sources
The same CVE in the primary databases and aggregators, to check against.
- CVE.org record (opens in a new tab)
- CVE List JSON (GitHub) (opens in a new tab)
- NVD (opens in a new tab)
- FIRST EPSS (opens in a new tab)
- OSV.dev (opens in a new tab)
- GitHub Advisory GHSA-f65r-h4g3-3h9h (opens in a new tab)
- CIRCL Vulnerability-Lookup (opens in a new tab)
Badge for a README or a ticket
[](https://trackcve.app/cve/CVE-2026-48797)Record last updated .