Cyber Posture

CVE-2025-26210

HighPublic PoC

Published: 03 September 2025

Published
03 September 2025
Modified
26 September 2025
KEV Added
Patch
CVSS Score 8.8 CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H
EPSS Score 0.0013 31.5th percentile
Risk Priority 18 60% EPSS · 20% KEV · 20% CVSS

Summary

CVE-2025-26210 is a high-severity Cross-site Scripting (CWE-79) vulnerability in Deepseek Deepseek-R1. Its CVSS base score is 8.8 (High).

Operationally, exploitation aligns with the MITRE ATT&CK technique Exploit Public-Facing Application (T1190); ranked at the 31.5th percentile by exploit likelihood (below the median); it is not currently listed in the CISA KEV catalog; a public proof-of-concept is referenced.

This vulnerability is AI-related — categorised as Enterprise AI Assistants; in the LLM/Generative AI Risks risk domain.

The strongest mitigations our analysis identified are NIST 800-53 SC-18 (Mobile Code) and SI-10 (Information Input Validation).

Threat & Defense at a Glance

What attackers do: exploitation maps to Exploit Public-Facing Application (T1190) and 7 other techniques. What defenders deploy: see the NIST 800-53 controls recommended below.
Threat & Defense Details

Mitigating Controls (NIST 800-53 r5)AI

prevent

SI-10 enforces validation and sanitization of user inputs to block malicious JavaScript payloads that enable XSS execution in the DeepSeek chat domain.

prevent

SI-15 requires output filtering and encoding to prevent injected JavaScript from executing in the victim's browser context as demonstrated in this CVE.

prevent

SC-18 restricts mobile code technologies like JavaScript, directly countering arbitrary script execution within the run-html-chat.deepseeksvc.com domain.

MITRE ATT&CK Enterprise TechniquesAI

T1190 Exploit Public-Facing Application Initial Access
Adversaries may attempt to exploit a weakness in an Internet-facing host or system to initially access a network.
T1203 Exploitation for Client Execution Execution
Adversaries may exploit software vulnerabilities in client applications to execute code.
T1059.007 JavaScript Execution
Adversaries may abuse various implementations of JavaScript for execution.
T1082 System Information Discovery Discovery
An adversary may attempt to get detailed information about the operating system and hardware, including version, patches, hotfixes, service packs, and architecture.
T1217 Browser Information Discovery Discovery
Adversaries may enumerate information about browsers to learn more about compromised environments.
T1539 Steal Web Session Cookie Credential Access
An adversary may steal web application or service session cookies and use them to gain access to web applications or Internet services as an authenticated user without needing credentials.
T1555.003 Credentials from Web Browsers Credential Access
Adversaries may acquire credentials from web browsers by reading files specific to the target browser.
T1567 Exfiltration Over Web Service Exfiltration
Adversaries may use an existing, legitimate external Web service to exfiltrate data rather than their primary command and control channel.
Why these techniques?

XSS enables exploitation of public-facing web app (T1190), client-side execution via JS (T1203, T1059.007), discovery of system/browser info (T1082, T1217), theft of web session cookies and browser credentials (T1539, T1555.003), and exfiltration to external web services via fetch.

NVD Description

DeepSeek R1 through V3.1 allows XSS, as demonstrated by JavaScript execution in the context of the run-html-chat.deepseeksvc.com domain. NOTE: some third parties have indicated that this is intended behavior.

Deeper analysisAI

CVE-2025-26210 is a cross-site scripting (XSS) vulnerability affecting DeepSeek R1 through V3.1. It enables JavaScript execution within the context of the run-html-chat.deepseeksvc.com domain, as demonstrated in public reports. The issue is classified under CWE-79 and carries a CVSS v3.1 base score of 8.8 (AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H), indicating high severity due to its network accessibility, low attack complexity, lack of required privileges, and significant impacts on confidentiality, integrity, and availability.

The vulnerability can be exploited by any network-accessible attacker with no authentication required, though it relies on user interaction such as clicking a malicious link or input. Successful exploitation allows arbitrary JavaScript execution in the victim's browser under the deepseeksvc.com domain context, potentially leading to session hijacking, data theft, account takeover, or further malicious actions within the affected service.

References include the official DeepSeek site (https://deepseek.com), a detailed HackMD post (https://hackmd.io/@MrqrFIlhQFi7vUwkqbrXDw/deepseek), and a YouTube demonstration (https://youtu.be/IgQwy52FVT4). Some third parties have noted that the behavior may be intended, with no specific patches or mitigations detailed in the available information.

Details

CWE(s)

Affected Products

deepseek
deepseek-r1
1.0
deepseek
deepseek-v2
all versions
deepseek
deepseek-v3
1.0

AI Security AnalysisAI

AI Category
Enterprise AI Assistants
Risk Domain
LLM/Generative AI Risks
OWASP Top 10 for LLMs 2025
None mapped
Classification Reason
DeepSeek is an AI chat platform (chat.deepseek.com) providing LLM-based code generation and execution features, fitting enterprise AI assistants; vulnerability exploits AI-generated insecure HTML/JS leading to XSS in the platform's runtime context.

CVEs Like This One

CVE-2025-23668Shared CWE-79
CVE-2025-5352Shared CWE-79
CVE-2024-7044Shared CWE-79
CVE-2025-68669Shared CWE-79
CVE-2025-0747Shared CWE-79
CVE-2024-54142Shared CWE-79
CVE-2024-13626Shared CWE-79
CVE-2025-23621Shared CWE-79
CVE-2026-34563Shared CWE-79
CVE-2026-1008Shared CWE-79

References