Adaptive Model Inversion Attacks Generalize a Privacy-Robustness Tradeoff

In this paper, we show that standard evaluations of high-resolution Model Inversion Attacks MIAs significantly underestimate training-data privacy leakage. State-of-the-art privacy defenses, standard training techniques such as MixUp and Adversarial Training, and undefended models all leak traini...

The Poisoned Conversation: Privacy-Leaking Watermarks in Unified Multimodal Models

Multimodal models are increasingly shifting toward unified architectures that understand and generate text, images, and other modalities within a shared conversational context. This design enables fluid interaction across modalities, but it also changes the privacy threat model: Information...

Privacy Leakage through AI-Mediated Analysis of Smartphone Data

Over the past thirty years, the online advertising industry built a large-scale data collection ecosystem, with the goal of tracking a user's online activity to infer their demographics and interests. Traditionally, the ecosystem relied upon the collation and analysis of highly-structured text da...

When Malicious Instructions Persist: Persistent Memory Poisoning Attack on Harness-Based Agents

Harness design has transformed the development of LLM-based agents by integrating memory, tool use, and runtime control. However, this design also introduces security and privacy risks because malicious instructions from external sources may be written into persistent memory and persist across...

Cascading Gradient Inversion Via LT-Code Inspired Peeling in Federated Learning

Federated learning shares model updates rather than raw data, yet these updates can be inverted to reconstruct the clients' training data. Analytic reconstruction attacks, which invert a gradient in closed form, degrade as the batch grows: prior single-round attacks recover only about half of a...

Hearing the Whispers: Black-Box Membership Inference Attacks on Finetuned TTS Models

Text-to-Speech TTS foundation models are increasingly fine-tuned on private datasets to synthesize highly personalized voices, introducing severe privacy risks by exposing both biometric identities and sensitive speech content. Existing black-box membership inference attacks MIAs follow a two-sta...

WeClawArena: An Auditable Sandbox and Benchmark for Cross-User Agents Collaboration and Security in Human-Centered Agent Networks

Recent advances in persistent personal-agent frameworks are making human-centered agent networks realistic deployment targets: each user can be served by an AI agent that acts on the user's behalf, maintains state, and communicates with other agents through social and task relations. In these...

When Agents Learn to Be You: Benchmarking Privacy Leakage, Impersonation Risk, and Defenses in Persona Skills

Persona skills distill personal interaction histories into portable and executable artifacts for downstream agents. While enabling flexible personalization, this process concentrates fragmented personal signals, amplifies their impact through reuse, and challenges defenses designed for individual...

CVE-2026-14223

The Easy Appointments WordPress plugin (versions before 3.12.28 ) is vulnerable to an Insecure Direct Object Reference (IDOR) . The root cause is a failure to verify ownership or capability when returning stored customer details. This allows users with subscriber-level access to read any customer...

USN-8569-1: Linux kernel (HWE) vulnerabilities

It was discovered that some AMD processors did not properly clear data in the floating point divider unit during speculative execution. A local attacker could use this to expose sensitive information. CVE-2025-54505 It was discovered that some AMD Zen 2 processors did not properly isolate shared...

GO-2026-5722 free5gc UDR nudr-dr influenceData/subs-to-notify leaks SUPI in error response body without authentication in github.com/free5gc/udr

free5gc UDR nudr-dr influenceData/subs-to-notify leaks SUPI in error response body without authentication in github.com/free5gc/udr...

CVE-2026-53675

BuddyPress 14.4.0 contains an insecure direct object reference vulnerability in the friends REST API that allows any authenticated attacker to enumerate another user's complete friend list. Attackers can query the friends endpoint with an arbitrary userid because the getitemspermissionscheck meth...

EUVD-2020-31249

HelloTalk through 3.4.1 stores full-precision GPS coordinates even when the user had intended to share only a country or city. Furthermore, these coordinates are placed into a database on the client of other users. The client side was changed in 2019 to encrypt that database...

Noisy Networks, Nosy Neighbors: Simple Privacy Attacks against Residential Wireless Traffic

Smart devices, such as light bulbs, TVs, fridges, etc., equipped with computing capabilities and wireless communication, are part of everyday life in many households. Previous work has already shown that a passive eavesdropper can derive private information, household routines, etc., from the...

HAPI FHIR HTTP authentication leak in redirects

When setting headers in HTTP requests, the internal HTTP client sends headers first to the host in the initial URL but also, if asked to follow redirects and a 30X HTTP response code is returned, to the host mentioned in URL in the Location: response header value. Sending the same set of headers ...

Aegis: Towards Governance, Integrity, and Security of AI Voice Agents

With the rapid advancement and adoption of Audio Large Language Models ALLMs, voice agents are now being deployed in high-stakes domains such as banking, customer service, and IT support. However, their vulnerabilities to adversarial misuse still remain unexplored. While prior work has examined...

CVE-2026-22246

Mastodon is a free, open-source social network server based on ActivityPub. Mastodon 4.3 added notifications of severed relationships, allowing end-users to inspect the relationships they lost as the result of a moderation action. The code allowing users to download lists of severed relationships...

A Practical Framework for Evaluating Medical AI Security: Reproducible Assessment of Jailbreaking and Privacy Vulnerabilities across Clinical Specialties

Medical Large Language Models LLMs are increasingly deployed for clinical decision support across diverse specialties, yet systematic evaluation of their robustness to adversarial misuse and privacy leakage remains inaccessible to most researchers. Existing security benchmarks require GPU cluster...

DualTAP: A Dual-Task Adversarial Protector for Mobile MLLM Agents

The reliance of mobile GUI agents on Multimodal Large Language Models MLLMs introduces a severe privacy vulnerability: screenshots containing Personally Identifiable Information PII are often sent to untrusted, third-party routers. These routers can exploit their own MLLMs to mine this data,...

LoRA-Leak: Membership Inference Attacks against LoRA Fine-Tuned Language Models

Language Models LMs typically adhere to a "pre-training and fine-tuning" paradigm, where a universal pre-trained model can be fine-tuned to cater to various specialized domains. Low-Rank Adaptation LoRA has gained the most widespread use in LM fine-tuning due to its lightweight computational cost...