Skip to content
Submit incident
AII-1000-25521
Well documentedThree or more independent publishers report the core facts.

Dify platform four critical vulnerabilities expose cross-tenant data

Researchers disclosed four vulnerabilities in the Dify open-source AI platform, including two critical flaws that required no authentication and enabled cross-tenant data exposure affecting over a million applications. The flaws allowed attackers to exfiltrate messages, access documents, and read chat histories across customer accounts without authorization.

23 June 2026 ยท Dify ยท Security and cyberattacks
23 June 2026
AII-1000-255214 sources โ†“
Quick facts
Category
Security and cyberattacks
Source
Security Affairs
Date
23 June 2026
Drafted by
Raidu Incident Detection

What happened

Zafran Labs researchers disclosed four vulnerabilities in Dify, an open-source AI platform used by major companies including Volvo and Maersk to run over a million applications across over 60 industries. Two vulnerabilities are of critical severity and two require no authentication at all.

CVE-2026-41947 (CVSS 9.1) in Dify's tracing system allows attackers who sign up as console users to configure tracing for any publicly accessible application and create a persistent exfiltration channel for messages and responses. CVE-2026-41948 (CVSS 9.4) in the Plugin Daemon contains path traversal and endpoint access flaws that require no login and allow access to arbitrary endpoints within the service.

CVE-2026-41949 and CVE-2026-41950 involve file access without ownership or tenant checks, allowing any console user to preview any document in the system or attach another user's file UUID to chat messages for an AI to read back. Dify also ran a PDFium binary vulnerable to CVE-2024-5846, disclosed in June 2024, for more than a year and a half until December 21, 2025.

Reported impact

The vulnerabilities enable cross-tenant data exposure on Dify's cloud service, meaning one customer's private data was readable by another customer. Attackers can exfiltrate all messages and responses sent in applications and access documents and conversations without authorization.

Classification

Category
Security and cyberattacks
Industry
IT infrastructure and hosting
Harm status
Potential harm
Intent
Unintentional
Timing
Post-deployment
Harmed
Consumers, Business
Failure modes
Data leak
AI systems
Dify

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

Security Affairs
DifyTap: Four Bugs Put over 1 million AI Apps at Risk - Security Affairs
23 June 2026
Dark Reading
DifyTap Bugs Let Attackers 'Wiretap' AI Chat Histories
23 June 2026
SC Media
4 vulnerabilities in Dify expose cross-tenant data
23 June 2026
SecurityBrief Asia
Dify flaws expose cross-tenant AI data, Zafran says
24 June 2026