CVE-2026-38743 Apache Airflow: Dags endpoint might provide access to otherwise inaccessible entities

The authenticated /ui/dags endpoint did not enforce per-DAG access control on embedded Human-in-the-Loop HITL and TaskInstance records: a logged-in Airflow user with read access to at least one DAG could retrieve HITL prompts including their request parameters and full TaskInstance details for DA...

CVE-2026-38743: Insufficient Granularity of Access Control

The authenticated /ui/dags endpoint did not enforce per-DAG access control on embedded Human-in-the-Loop HITL and TaskInstance records: a logged-in Airflow user with read access to at least one DAG could retrieve HITL prompts including their request parameters and full TaskInstance details for DA...

CVE-2026-41271

Flowise is a drag & drop user interface to build a customized large language model flow. Prior to 3.1.0, a Server-Side Request Forgery SSRF vulnerability exists in FlowiseAI's POST/GET API Chain components that allows unauthenticated attackers to force the server to make arbitrary HTTP requests t...

CVE-2026-6848 Quay: red hat quay: authentication bypass allows privileged actions without valid credentials

A flaw was found in Red Hat Quay. When Red Hat Quay requests password re-verification for sensitive operations, such as token generation or robot account creation, the re-authentication prompt can be bypassed. This allows a user with a timed-out session, or an attacker with access to an idle...

CVE-2026-41300

OpenClaw npm package OpenClaw (openclaw) before 2026.3.31 is affected by a trust-decline vulnerability that allows attacker-discovered endpoints to survive remote onboarding flows, enabling routing of gateway credentials to malicious endpoints. Affected versions are = 2026.3.31. If exploitation d...

CVE-2026-40525

OpenViking prior to version 0.3.9 contains an authentication bypass vulnerability in the VikingBot OpenAPI HTTP route surface where the authentication check fails open when the apikey configuration value is unset or empty. Remote attackers with network access to the exposed service can invoke...

ARES: Adaptive Red-Teaming and End-To-End Repair of Policy-Reward System

Reinforcement Learning from Human Feedback RLHF is central to aligning Large Language Models LLMs, yet it introduces a critical vulnerability: an imperfect Reward Model RM can become a single point of failure when it fails to penalize unsafe behaviors. While existing red-teaming approaches...

GuardPhish: Securing Open-Source LLMs from Phishing Abuse

The rapid adoption of open-source Large Language Models LLMs in offline and enterprise environments has introduced a largely unexamined security risk like susceptibility to adversarial phishing prompts under static safety configurations. In this work, we systematically investigate this...

Incomplete List of Disallowed Inputs

Overview flowise-ui is a Affected versions of this package are vulnerable to Incomplete List of Disallowed Inputs in the run method of the AirtableAgents class, which evaluates LLM-generated Python scripts in a non-sandboxed environment. An attacker can execute arbitrary code on the server by...

GHSA-JGQ2-VQ69-GR6H OpenViking: Unauthenticated remote bot control via OpenAPI HTTP routes

OpenViking prior to commit c7bb167 contains an authentication bypass vulnerability in the VikingBot OpenAPI HTTP route surface where the authentication check fails open when the apikey configuration value is unset or empty. Remote attackers with network access to the exposed service can invoke...

PT-2026-33478

Name of the Vulnerable Software and Affected Versions OpenViking versions prior to commit c7bb167 Description An authentication bypass exists in the VikingBot OpenAPI HTTP route surface. The issue occurs when the api key configuration value is unset or empty, causing the authentication check to...

SUSE CVE-2026-40505

MuPDF before 1.27 contains an ANSI injection vulnerability in mutool that allows attackers to inject arbitrary ANSI escape sequences through crafted PDF metadata fields. Attackers can embed malicious ANSI escape codes in PDF metadata that are passed unsanitized to terminal output when running...

Paperclip: Unauthenticated Access to Multiple API Endpoints in Authenticated Mode

Summary Several API endpoints in authenticated mode have no authentication at all. They respond to completely unauthenticated requests with sensitive data or allow state-changing operations. No account, no session, no API key needed. Verified against the latest version. Discord: sagi03581 Steps t...

Server-side Request Forgery (SSRF)

Overview flowise-components is a Flowiseai Components Affected versions of this package are vulnerable to Server-side Request Forgery SSRF via postCore.ts. An attacker can cause the server to make arbitrary HTTP requests to internal or external systems by injecting malicious prompt templates that...

CLSA-2026-1768300651 git: Fix of CVE-2024-50349

CVE-2024-50349: fix ANSI escape sequence vulnerability that occurs when asking for credentials interactively...

CVE-2026-40505

A flaw was found in MuPDF mutool. This vulnerability allows a local attacker to embed malicious ANSI escape sequences within a PDF's metadata. When a user views the PDF's information using the mutool info command, these unsanitized sequences are processed by the terminal. This can lead to the...

PT-2026-33222

Name of the Vulnerable Software and Affected Versions MuPDF mutool affected versions not specified Description MuPDF mutool fails to sanitize PDF metadata fields before writing them to terminal output. This allows the injection of arbitrary ANSI escape sequences—codes used to control terminal...

JLSEC-2026-100 Deno is vulnerable to race condition via interactive permission prompt spoofing

Impact Multi-threaded programs were able to spoof interactive permission prompt by rewriting the prompt to suggest that program is waiting on user confirmation to unrelated action. A malicious program could clear the terminal screen after permission prompt was shown and write a generic message li...

Open WebUI 安全漏洞

Open WebUI is an extensible, feature-rich, and user-friendly self-hosted WebUI under open source. Versions of Open WebUI 0.7.2 and earlier contain security vulnerabilities. These vulnerabilities stem from a feature that allows for image editing through prompts, which enables blind server-side...

Honeypot Protocol

Trusted monitoring, the standard defense in AI control, is vulnerable to adaptive attacks, collusion, and strategic attack selection. All of these exploit the fact that monitoring is passive: it observes model behavior but never probes whether the model would behave differently under different...