CVE-2026-41279: Flowise unauth API key abuse

CISO Take

Flowise's text-to-speech endpoint is whitelisted without authentication and blindly trusts a caller-supplied credentialId to decrypt and invoke stored AI provider keys (OpenAI, ElevenLabs) — any unauthenticated party on the internet can trigger this with zero privileges. A public PoC already exists and EPSS places this in the top 83rd percentile for exploitation likelihood, meaning active abuse is plausible right now. The blast radius is financial and operational: attackers can silently drain AI API credits, exhaust provider rate limits, and disrupt production AI workflows without leaving any trace in Flowise auth logs since no login ever occurs. Upgrade to Flowise 3.1.0 immediately, rotate every stored third-party API credential as a precaution, and if patching is blocked, firewall POST /api/v1/text-to-speech/generate at the edge.

Sources: NVD EPSS GitHub Advisory ATLAS

What is the risk?

High. Network-accessible with no authentication, no user interaction, and low attack complexity — combined with a public PoC and EPSS top-83rd-percentile ranking, this is effectively trivial for any attacker who can reach a Flowise instance. Flowise carries 58 prior CVEs in the same package, signaling systemic security debt rather than an isolated miss. Organizations running internet-facing Flowise deployments with stored AI provider credentials face immediate financial drain and potential service disruption with no authentication barrier in the way.

How does the attack unfold?

Initial Access

Adversary identifies a publicly accessible Flowise instance via internet scanning and confirms it is running a pre-3.1.0 version through UI or API version disclosure.

AML.T0049

Credential ID Enumeration

Adversary iterates through credentialId values via the unauthenticated TTS endpoint using IDOR techniques, confirming which IDs map to valid stored AI provider credentials.

AML.T0055

Exploitation

Adversary calls POST /api/v1/text-to-speech/generate with a valid credentialId, causing the server to server-side decrypt and use stored API keys (OpenAI, ElevenLabs) with no authentication event logged.

AML.T0083

Impact

Adversary sends bulk automated TTS requests, draining victim AI API credits, exhausting provider rate limits, and disrupting legitimate AI-dependent services at the victim's financial cost.

AML.T0034

Initial Access

Adversary identifies a publicly accessible Flowise instance via internet scanning and confirms it is running a pre-3.1.0 version through UI or API version disclosure.

AML.T0049

Credential ID Enumeration

Adversary iterates through credentialId values via the unauthenticated TTS endpoint using IDOR techniques, confirming which IDs map to valid stored AI provider credentials.

AML.T0055

Exploitation

Adversary calls POST /api/v1/text-to-speech/generate with a valid credentialId, causing the server to server-side decrypt and use stored API keys (OpenAI, ElevenLabs) with no authentication event logged.

AML.T0083

Impact

Adversary sends bulk automated TTS requests, draining victim AI API credits, exhausting provider rate limits, and disrupting legitimate AI-dependent services at the victim's financial cost.

AML.T0034

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
Flowise	npm	—	No patch

Do you use Flowise? You're affected.

How severe is it?

CVSS 3.1

7.5 / 10

EPSS

0.3%

chance of exploitation in 30 days

Higher than 17% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

Exploit Available

Exploitation: MEDIUM

Sophistication

Trivial

Exploitation Confidence

medium

○ Public PoC indexed (trickest/cve)

Composite signal derived from CISA KEV, VulnCheck KEV, CISA SSVC, EPSS, Metasploit, Exploit-DB, trickest/cve, Nuclei templates, and inthewild.io exploitation reports.

What is the attack surface?

AV Network

AC Low

PR None

UI None

S Unchanged

C None

I None

A High

What should I do?

5 steps

Upgrade to Flowise 3.1.0 immediately — the patch adds authentication to the TTS endpoint.
Rotate all stored third-party API credentials (OpenAI, ElevenLabs, etc.) in Flowise as a precaution, even post-patch, since there is no way to audit whether keys were abused prior to patching.
If immediate upgrade is blocked, restrict external access to POST /api/v1/text-to-speech/generate at your WAF or reverse proxy — allow only trusted internal IPs.
Enable spend alerts and usage anomaly notifications at your AI providers (OpenAI dashboard, ElevenLabs billing) to detect unauthorized TTS consumption.
Review Flowise audit logs for unexpected credentialId references and cross-reference with provider API call timestamps.

What does CISA's SSVC say?

Decision Track

Exploitation none

Automatable No

Technical Impact partial

Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.

How is it classified?

Auth Bypass DoS Data Extraction Agent API Framework AML.T0034 - Cost Harvesting AML.T0034.000 - Excessive Queries AML.T0049 - Exploit Public-Facing Application AML.T0055 - Unsecured Credentials AML.T0083 - Credentials from AI Agent Configuration

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Article 15 - Accuracy, robustness and cybersecurity Article 9 - Risk management system

ISO 42001

A.7 - AI system design and development Clause 6.1 - Actions to address risks and opportunities

NIST AI RMF

GOVERN 1.1 - Policies and procedures for AI risk management MANAGE 2.2 - Risk treatment mechanisms for AI systems

OWASP LLM Top 10

LLM06:2025 - Excessive Agency LLM10:2025 - Unbounded Consumption

Frequently Asked Questions

What is CVE-2026-41279?

Flowise's text-to-speech endpoint is whitelisted without authentication and blindly trusts a caller-supplied credentialId to decrypt and invoke stored AI provider keys (OpenAI, ElevenLabs) — any unauthenticated party on the internet can trigger this with zero privileges. A public PoC already exists and EPSS places this in the top 83rd percentile for exploitation likelihood, meaning active abuse is plausible right now. The blast radius is financial and operational: attackers can silently drain AI API credits, exhaust provider rate limits, and disrupt production AI workflows without leaving any trace in Flowise auth logs since no login ever occurs. Upgrade to Flowise 3.1.0 immediately, rotate every stored third-party API credential as a precaution, and if patching is blocked, firewall POST /api/v1/text-to-speech/generate at the edge.

Is CVE-2026-41279 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2026-41279, increasing the risk of exploitation.

How to fix CVE-2026-41279?

1. Upgrade to Flowise 3.1.0 immediately — the patch adds authentication to the TTS endpoint. 2. Rotate all stored third-party API credentials (OpenAI, ElevenLabs, etc.) in Flowise as a precaution, even post-patch, since there is no way to audit whether keys were abused prior to patching. 3. If immediate upgrade is blocked, restrict external access to POST /api/v1/text-to-speech/generate at your WAF or reverse proxy — allow only trusted internal IPs. 4. Enable spend alerts and usage anomaly notifications at your AI providers (OpenAI dashboard, ElevenLabs billing) to detect unauthorized TTS consumption. 5. Review Flowise audit logs for unexpected credentialId references and cross-reference with provider API call timestamps.

What systems are affected by CVE-2026-41279?

This vulnerability affects the following AI/ML architecture patterns: AI agent frameworks, LLM workflow orchestration, No-code/low-code LLM builders, Multi-tenant AI platforms.

What is the CVSS score for CVE-2026-41279?

CVE-2026-41279 has a CVSS v3.1 base score of 7.5 (HIGH). The EPSS exploitation probability is 0.26%.

What is the AI security impact?

Affected AI Architectures

AI agent frameworksLLM workflow orchestrationNo-code/low-code LLM buildersMulti-tenant AI platforms

MITRE ATLAS Techniques

AML.T0034 Cost Harvesting

AML.T0034.000 Excessive Queries

AML.T0049 Exploit Public-Facing Application

AML.T0055 Unsecured Credentials

AML.T0083 Credentials from AI Agent Configuration

Compliance Controls Affected

EU AI Act: Article 15, Article 9

ISO 42001: A.7, Clause 6.1

NIST AI RMF: GOVERN 1.1, MANAGE 2.2

OWASP LLM Top 10: LLM06:2025, LLM10:2025

What are the technical details?

Original Advisory

Flowise is a drag & drop user interface to build a customized large language model flow. Prior to 3.1.0, the text-to-speech generation endpoint (POST /api/v1/text-to-speech/generate) is whitelisted (no auth) and accepts a credentialId directly in the request body. When called without a chatflowId, the endpoint uses the provided credentialId to decrypt the stored credential (e.g., OpenAI or ElevenLabs API key) and generate speech. This vulnerability is fixed in 3.1.0.

Exploitation Scenario

An adversary scans the internet for publicly accessible Flowise instances — version disclosure is often available via UI headers or API responses. With the target confirmed as pre-3.1.0, they call POST /api/v1/text-to-speech/generate without any session token, iterating through credentialId values (sequential integers or UUIDs enumerable via IDOR). Each successful hit causes the Flowise server to decrypt and use the stored API key server-side to generate TTS output. The adversary automates bulk requests to exhaust OpenAI or ElevenLabs credits, potentially reselling TTS capacity or simply destroying availability of AI features for legitimate users — all while leaving zero footprint in authentication logs since no login ever occurred.

Weaknesses (CWE)

CWE-639 Authorization Bypass Through User-Controlled Key Primary

CWE-639 — Authorization Bypass Through User-Controlled Key: The system's authorization functionality does not prevent one user from gaining access to another user's data or record by modifying the key value identifying the data.

[Architecture and Design] For each and every data access, ensure that the user has sufficient privilege to access the record that is being requested.
[Architecture and Design, Implementation] Make sure that the key that is used in the lookup of a specific user's record is not controllable externally by the user or that any tampering can be detected.

Source: MITRE CWE corpus.