Personal AI agents built on large language models (LLMs) are increasingly given access to a user's private data and communications in order to provide personalized assistance. This access creates a persistent privacy risk: the agent must decide whether a given sensitive information should be disclosed to a particular p...
A targeted prompt-level defense is evaluated and finds that it can reduce memory injection in many settings, but provides limited protection once the persistent memory has been poisoned.
Shu-Huai Huang, Jing-Feng Zhang, Hong Jia· 1 citation
JavaScript powers approximately 98.8% of all websites, making vulnerabilities in its code a significant security risk, yet existing detection approaches such as Static Application Security Testing (SAST) tools often fail to identify many real-world vulnerabilities when applied to isolated code snippets. This paper pres...
Manit Kaushik, Ishir Bhardwaj, Pranav Gupta et al.· 0 citations
TyPatch is presented, which decouples patch-specific defect semantics from analyzer implementation and uses 88.3-90.1% fewer generation tokens than the state-of-the-art complete-checker construction workflow, while its initial report pools achieve 3.95 times the precision of those produced by that workflow.
Ruo-Yu Wang, Tuo Li, Jia Li· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
Large language models are increasingly benchmarked against classical machine learning for network intrusion detection (NIDS), almost always using same-dataset evaluation, and that protocol turns out to be incomplete. Evaluating XGBoost and RoBERTa-LoRA on two independently collected NetFlow v2 networks across three axe...
Muhammad Ebad Atif, Muhammad Haider Ali· 0 citations
A protocol-aware empirical audit is introduced in which the server commits to a single shared candidate bank and replaces roughly 1% of its entries with probes derived from a known, non-private canary, to quantify the gap between formal worst-case privacy and leakage achievable through protocol-valid candidate-bank man...
Sai Aparna Aketi, Enayat Ullah, Shripad Gade· 0 citations
SkillAtlas, a hosted attack trace library that converts private agent-skill security report bundles into reviewed, redacted, and searchable public cases, is presented.
Yu-Xin Tian, Zenghao Duan, Liang Pang et al.· 0 citations
By acquiring compute, credentials, accounts, services, and other agents, autonomous AI agents can introduce new authority into a task. Payment, budget, OAuth, mandate, and fulfillment checks can validate transaction conditions without deciding whether a returned resource may become usable authority. This post-fulfillme...
ADAE is introduced, a proposed AI engineering subdiscipline concerned with establishing measurable, continuous, and actionable accountability for deployed AI systems and treats accountability as a deployment-layer property rather than solely as a property of an individual model.
It is found that user studies explain substantially more variation in mechanism acceptance than physical deviation, although physical deviation remains significant.
Sensitive attributes such as age, income, and occupation can be inferred from user-generated content by aggregating indirect cues across many ordinary posts. LLM-based profilers can perform this aggregation automatically and with high accuracy, which makes large-scale personal attribute inference a major privacy threat...
Ahmed Sohair Khan, Estrid He, Chenglong Ma et al.· 0 citations
