Researchers have developed AirflowAttack, a novel method to create adversarial perturbations for vision-language models (VLMs) used in infrared remote sensing. This attack weaponizes thermal-airflow turbulence, generating physically plausible perturbations that can reduce scene classification accuracy by up to 38.2% across various state-of-the-art VLMs. Notably, the attack can even cause some models to misinterpret the perturbations as genuine thermal evidence, highlighting significant vulnerabilities in the deployment of these models in security-critical applications. AI
IMPACT Exposes critical vulnerabilities in infrared remote-sensing VLMs, potentially impacting their deployment in security-sensitive applications.
RANK_REASON Academic paper detailing a new adversarial attack method. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Hugging Face Daily Papers →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
