Token by Token, Compromised: Backdoor Vulnerabilities in Unified Autoregressive Models

Unified autoregressive models UAMs are transformer models that generate text as well as image tokens within a single autoregressive pass. Shared parameters and a multimodal vocabulary simplify the training pipeline and facilitate flexible multimodal generation, yet might introduce new...

Babel: Jailbreaking Safety Attention Via Obfuscation Distribution Optimized Sampling

Despite rigorous safety alignment, Large Language Models LLMs remain vulnerable to jailbreak attacks. Existing black-box methods often rely on heuristic templates or exhaustive trials, lacking mechanistic interpretability and query efficiency. In this study, we investigate an intrinsic...

Backdooring Masked Diffusion Language Models

Masked diffusion language models MDLMs are emerging as a compelling new paradigm for text generation, but their training-time security remains largely unexplored. Existing backdoor attacks on Gaussian diffusion models or autoregressive language models do not directly apply to MDLMs because MDLMs...

CVE-2026-8755

A flaw has been found in fishaudio Bert-VITS2 up to 8f7fbd8c4770965225d258db548da27dc8dd934c. The affected element is the function getallmodels of the file hiyoriUI.py of the component Model Handler. This manipulation causes path traversal. The attack can be initiated remotely. The exploit has be...

CVE-2026-8755

A flaw has been found in fishaudio Bert-VITS2 up to 8f7fbd8c4770965225d258db548da27dc8dd934c. The affected element is the function getallmodels of the file hiyoriUI.py of the component Model Handler. This manipulation causes path traversal. The attack can be initiated remotely. The exploit has be...

CVE-2026-8755 fishaudio Bert-VITS2 Model hiyoriUI.py _get_all_models path traversal

A flaw has been found in fishaudio Bert-VITS2 up to 8f7fbd8c4770965225d258db548da27dc8dd934c. The affected element is the function getallmodels of the file hiyoriUI.py of the component Model Handler. This manipulation causes path traversal. The attack can be initiated remotely. The exploit has be...

Rethinking Side-Channel Analysis: Automated Discovery and Analysis of Side-Channel Leakage with LLM-Assisted Agents

Side-channel attacks exploit unintended information leakage from system behavior and continue to pose serious privacy risks in modern platforms. Despite extensive prior work, side-channel analysis remains largely manual and fragmented, typically assuming predefined target events and a fixed set o...

PT-2026-41566

Name of the Vulnerable Software and Affected Versions fishaudio Bert-VITS2 versions prior to 8f7fbd8c4770965225d258db548da27dc8dd934c Description A path traversal flaw exists in the Model Handler component, specifically within the get all models function of the hiyoriUI.py file. This issue allows...

Bert-VITS2 路径遍历漏洞

Bert-VITS2 is a core text-to-speech model developed by Fish Audio. Bert-VITS2 has a path traversal vulnerability, which stems from the getallmodels function in the hiyoriUI.py file within the Model Handler component. Attackers could potentially exploit this vulnerability remotely...

A Red Teaming Framework for Evaluating Robustness of AI-Enabled Security Orchestration, Automation, and Response Systems

AI-enabled Security Orchestration, Automation, and Response SOAR systems increasingly employ autonomous agents for cyber defense, yet their resilience to adaptive adversaries is underexplored. We introduce an autonomous red teaming framework that integrates large language models LLMs with...

CVE-2026-45365

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.8.11, an internal-only bypassfilter parameter is exposed on the /openai/chat/completions and /ollama/api/chat HTTP endpoints via FastAPI query string binding, allowing any authenticated...

CVE-2026-45345

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.5.7, a user can modify another user's model even if its visibility is set to Private. By changing the access permissions during editing, unauthorized access can be gained. This...

EUVD-2026-30650

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.5.7, a user can modify another user's model even if its visibility is set to Private. By changing the access permissions during editing, unauthorized access can be gained. This...

CVE-2026-45345 Open WebUI: Missing authorization check at the model update function - models from other users can be updated

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.5.7, a user can modify another user's model even if its visibility is set to Private. By changing the access permissions during editing, unauthorized access can be gained. This...

CVE-2026-45345 Open WebUI: Missing authorization check at the model update function - models from other users can be updated

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.5.7, a user can modify another user's model even if its visibility is set to Private. By changing the access permissions during editing, unauthorized access can be gained. This...

CVE-2026-45345

Open WebUI (self-hosted AI platform) has a vulnerability in the model update function prior to version 0.5.7 where an attacker could modify another user’s private model by changing access permissions during editing. The issue is confirmed in multiple sources (CVE-2026-45345, GHSA-gm54-m39w-grjp, ...

CVE-2026-45351 Open WebUI: Exposure of System Prompt to Regular User [Non-Admin]

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.8.9, when a regular user non-admin logs into the application, a http://IP:8080/api/models? web request is initiated by the application and in response, it reveals the system prompt of...

EUVD-2026-30646

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.8.9, when a regular user non-admin logs into the application, a http://IP:8080/api/models? web request is initiated by the application and in response, it reveals the system prompt of...

CVE-2026-45351

Open WebUI vulnerability CVE-2026-45351: A non-admin user could trigger a request to /api/models? and receive the system prompt of available models, revealing admin-set backend prompts and compromising confidentiality. This affects Open WebUI self-hosted offline AI platform versions prior to 0.8....

CVE-2026-45351 Open WebUI: Exposure of System Prompt to Regular User [Non-Admin]

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.8.9, when a regular user non-admin logs into the application, a http://IP:8080/api/models? web request is initiated by the application and in response, it reveals the system prompt of...