Signal is active with 2 mentions in latest observed window
Recommended action window: Monitor and triage in normal cycle
NVD description
AnythingLLM is an application that turns pieces of content into context that any LLM can use as references during chatting. In 1.11.1 and earlier, AnythingLLM Desktop contains a Streaming Phase XSS vulnerability in the chat rendering pipeline that escalates to Remote Code Execution on the host OS due to insecure Electron configuration. This works with default settings and requires no user interaction beyond normal chat usage. The custom markdown-it image renderer in frontend/src/utils/chat/markdown.js interpolates token.content directly into the alt attribute without HTML entity escaping. The PromptReply component renders this output via dangerouslySetInnerHTML without DOMPurify sanitization — unlike HistoricalMessage which correctly applies DOMPurify.sanitize().
Priority
LOW
Exploitation
NONE
PoC
YES
Patch
AVAILABLE
Momentum
STABLE
If you run products in this scope, you should treat this CVE as relevant to your environment.
| Date | Total | Labels |
|---|
| 2026-03-13 | 2 | Disclosure1General1 |
| 2026-03-16 | 1 | Disclosure1 |
| 2026-03-22 | 2 | Disclosure1Patch1 |
Critical: AnythingLLM Desktop XSS-to-RCE via insecure Electron config. Poisoned RAG documents or compromised LLM endpoints can achieve full host compromise. CVE-2026-32626, CVSS 9.6. Patch available. https://raxe.ai/labs/advisories/RAXE-2026-038
Post summary
The advisory discloses a critical XSS‑to‑RCE flaw in AnythingLLM Desktop (CVE‑2026‑32626) with CVSS 9.6 and confirms a patch is available.
CVE-2026-32626 is a perfect example of why AI tooling is becoming prime attack surface. Poisoned documents in your RAG pipeline achieving full host compromise through Electron misconfig - attackers don't need to break the model, just feed it. Monitor AI CVEs: https://vulntracker.io
Post summary
The post announces CVE‑2026‑32626 as a new vulnerability that enables attackers to compromise hosts via Electron misconfiguration by feeding poisoned documents into AI pipelines, but it gives no PoC, exploit, or patch details and does not confirm active exploitation.
[CVE-2026-32626: CRITICAL] AnythingLLM application poses a serious cyber threat due to an XSS vulnerability enabling Remote Code Execution. The flaw in the Electron configuration allows exploitation through ...#cve,CVE-2026-32626,#cybersecurity https://cvefind.com/CVE-2026-32626
Post summary
The post highlights CVE-2026-32626 as a critical XSS‑to‑RCE flaw in AnythingLLM, but offers no PoC, exploit code, patch information, or evidence of active exploitation.
CVE-2026-32626 AnythingLLM is an application that turns pieces of content into context that any LLM can use as references during chatting. In 1.11.1 and earlier, AnythingLLM Desktop… https://www.cve.org/CVERecord?id=CVE-2026-32626
Post summary
The article merely references CVE-2026-32626 in relation to AnythingLLM, with no further technical, exploit or mitigation details.
🚨 CVE-2026-32626: AnythingLLM has a Streaming Phas... LLM response injection bypasses DOMPurify in PromptReply component, weaponizing markdown-it's unescaped alt attributes ... https://zerodaysignal.com/vulnerability/CVE-2026-32626 #netsec #vulnerability #CVE #sysadmin #zeroday
Post summary
A newly disclosed CVE-2026-32626 in AnythingLLM exposes a response injection flaw bypassing DOMPurify, leveraging unescaped alt attributes in markdown‑it; no exploit code or patch details are provided.
1 of 1 entries
| Part | Vendor | Product | Version | Target SW | Target HW |
|---|---|---|---|---|---|
| App | mintplexlabs | anythingllm | - | - | - |