LLM-Guided Evolution Attacks Expose Vulnerabilities in Perceptual Hash Algorithms

Jul 14, 2026

A new preprint introduces an evolutionary framework, guided by large language models (LLMs), to generate targeted black-box attacks on widely used perceptual hash algorithms (PHAs) such as pHash, PDQ, PhotoDNA, and NeuralHash. The approach achieves comparable or superior attack success rates with fewer queries and lower image distortion than existing black-box methods, revealing previously unreported vulnerabilities in these content moderation tools.

Why it matters: The findings highlight significant security weaknesses in commonly deployed perceptual hash algorithms, underscoring the urgent need for more robust content moderation technologies.

Full story at: arXiv Cryptography and Security ↗