AI Adversarial Robustness

2 summarised stories about AI Adversarial Robustness, each linking back to the original source. Browse all topics →

Friday, 17 July 2026

CluCERT: Certifying LLM Robustness via Clustering-Guided Denoising Smoothing

arXiv cs.AI 6 hours ago

Researchers introduced CluCERT, a framework for certifying Large Language Model robustness against adversarial attacks using clustering-guided denoising smoothing techniques. The method achieved tighter certified robustness bounds compared to existing approaches while reducing computational costs through semantic clustering filters and accelerated synonym substitution strategies. CluCERT enables more efficient evaluation of LLM vulnerability to meaning-preserving input perturbations across various tasks and jailbreak defense scenarios.

AI Adversarial Robustness AI Robustness Certification AI Security Large Language Models

Large Audio Language Models for Spoofing-Aware Speaker Verification

arXiv cs.AI 6 hours ago

Researchers evaluated large audio language models for spoofing-aware speaker verification, a security task that distinguishes genuine voices from synthetic clones or spoofed audio. The models achieved near-chance performance in zero-shot settings but reached competitive results after task-specific adaptation and optimization techniques. Large audio language models can provide interpretable rationales for verification decisions while unifying detection of both spoofed audio and genuine speaker identification.

AI Adaptation & Learning AI Adversarial Robustness AI & Audio Processing AI & Cybersecurity