PT-2025-38953

Name of the Vulnerable Software and Affected Versions Azizul Hasan Text To Speech TTS Accessibility versions through 1.9.20 Description A missing authorization issue exists in Azizul Hasan Text To Speech TTS Accessibility. The issue involves exploiting incorrectly configured access control securi...

PT-2025-38801

Name of the Vulnerable Software and Affected Versions ken107 SiteNarrator Text-to-Speech Widget versions through 1.9 Description The ken107 SiteNarrator Text-to-Speech Widget contains a flaw due to improper neutralization of input during web page generation, leading to a Stored Cross-site Scripti...

Invisible Ears at Your Fingertips: Acoustic Eavesdropping Via Mouse Sensors

Modern optical mouse sensors, with their advanced precision and high responsiveness, possess an often overlooked vulnerability: they can be exploited for side-channel attacks. This paper introduces Mic-E-Mouse, the first-ever side-channel attack that targets high-performance optical mouse sensors...

GHSA-RCV9-QM8P-9P6J Hugging Face Transformers library has Regular Expression Denial of Service

A Regular Expression Denial of Service ReDoS vulnerability was discovered in the Hugging Face Transformers library, specifically within the normalizenumbers method of the EnglishNormalizer class. This vulnerability affects versions up to 4.52.4 and is fixed in version 4.53.0. The issue arises fro...

CVE-2025-6051

A Regular Expression Denial of Service ReDoS vulnerability was discovered in the Hugging Face Transformers library, specifically within the normalizenumbers method of the EnglishNormalizer class. This vulnerability affects versions up to 4.52.4 and is fixed in version 4.53.0. The issue arises fro...

CVE-2025-6051

CVE-2025-6051 is a ReDoS in Hugging Face Transformers’ EnglishNormalizer.normalize_numbers(), affecting versions up to 4.52.4 and fixed in 4.53.0. The issue arises from numeric string handling, enabling crafted inputs with long digit sequences to cause excessive CPU usage, impacting text-to-speec...

ENJ: Optimizing Noise with Genetic Algorithms to Jailbreak LSMs

The widespread application of Large Speech Models LSMs has made their security risks increasingly prominent. Traditional speech adversarial attack methods face challenges in balancing effectiveness and stealth. This paper proposes Evolutionary Noise Jailbreak ENJ, which utilizes a genetic algorit...

Security Bulletin: IBM Watson Speech Services Cartridge is vulnerable to an Uncontrolled Recursion vulnerability in Apache Commons [CVE-2025-48924]

Summary IBM Watson Speech Services Cartridge is vulnerable to an Uncontrolled Recursion vulnerability in Apache Commons Lang, caused by the methods ClassUtils.getClass... possibily throwing a StackOverflowError on very long inputs CVE-2025-48924. Apache Commons is used in our speech service...

Security Bulletin: IBM Watson Speech Services Cartridge is vulnerable to a path traversal vulnerability in rsync [CVE-2024-12088]

Summary IBM Watson Speech Services Cartridge is vulnerable to a path traversal vulnerability in rsync, due to an issue when using the --safe-links option, where the client fails to properly verify if a symbolic link destination sent from the server contains another symbolic link within it...

Security Bulletin: IBM Watson Speech Services Cartridge is vulnerable to an Improper Input Validation in Apache Tomcat [CVE-2025-31650]

Summary IBM Watson Speech Services Cartridge is vulnerable to an Improper Input Validation in Apache Tomcat, caused by incorrect error handling for some invalid HTTP priority headers, resulting in incomplete clean-up of the failed request, which creates a memory leak CVE-2025-31650 Apache Tomcat ...

Security Bulletin: IBM Watson Speech Services Cartridge is vulnerable to an Improper Input Validation in Spring [CVE-2025-22235]

Summary IBM Watson Speech Services Cartridge is vulnerable to an Improper Input Validation in Spring , caused by Spring Boot EndpointRequest.to creating the wrong matcher if the actuator endpoint is not exposed CVE-2025-22235 . Spring is used as part of our Java Microservices. This vulnerabilitiy...

Security Bulletin: IBM Watson Speech Services Cartridge is vulnerable to a Remote Command Execution in PyTorch [CVE-2025-32434]

Summary IBM Watson Speech Services Cartridge is vulnerable to a Remote Command Execution in PyTorch, due to a condition that exists when loading a model using torch.load with weightsonly=True CVE-2025-32434. PyTorch is used in our speech service runtimes. This vulnerabilitiy has been addressed...

ScamAgents: How AI Agents Can Simulate Human-Level Scam Calls

Large Language Models LLMs have demonstrated impressive fluency and reasoning capabilities, but their potential for misuse has raised growing concern. In this paper, we present ScamAgent, an autonomous multi-turn agent built on top of LLMs, capable of generating highly realistic scam call scripts...

Multilingual Source Tracing of Speech Deepfakes: a First Benchmark

Recent progress in generative AI has made it increasingly easy to create natural-sounding deepfake speech from just a few seconds of audio. While these tools support helpful applications, they also raise serious concerns by making it possible to generate convincing fake speech in many languages...

The Age-Checked Internet Has Arrived

Starting today, UK adults will have to prove their age to access porn online. Experts warn that a global wave of age-check laws threatens to chill speech and ultimately harm children and adults alike...

LENS-DF: Deepfake Detection and Temporal Localization for Long-Form Noisy Speech

This study introduces LENS-DF, a novel and comprehensive recipe for training and evaluating audio deepfake detection and temporal localization under complicated and realistic audio conditions. The generation part of the recipe outputs audios from the input dataset with several critical...

Frame-Level Temporal Difference Learning for Partial Deepfake Speech Detection

Detecting partial deepfake speech is essential due to its potential for subtle misinformation. However, existing methods depend on costly frame-level annotations during training, limiting real-world scalability. Also, they focus on detecting transition artifacts between bonafide and deepfake...

Exploiting Context-Dependent Duration Features for Voice Anonymization Attack Systems

The temporal dynamics of speech, encompassing variations in rhythm, intonation, and speaking rate, contain important and unique information about speaker identity. This paper proposes a new method for representing speaker characteristics by extracting context-dependent duration embeddings from...

CVE-2025-49836

GPT-SoVITS-WebUI is a voice conversion and text-to-speech webUI. In versions 20250228v3 and prior, there is a command injection vulnerability in webui.py changelabel function. pathlist takes user input, which is passed to the changelabel function, which concatenates the user input into a command...

CVE-2025-49840

GPT-SoVITS-WebUI is a voice conversion and text-to-speech webUI. In versions 20250228v3 and prior, there is an unsafe deserialization vulnerability in inferencewebui.py. The GPTdropdown variable takes user input and passes it to the changegptweights function. In changegptweights, the user input,...