Vision-Language Models (VLMs) are increasingly deployed in real-world applications, but their high inference cost makes them vulnerable to resource consumption attacks. Prior attacks attempt to extend VLM output sequences by optimizing adversarial images, thereby increasing inference costs. However, these extended outputs often introduce irrelevant abnormal content, compromising attack stealthiness. This trade-off between effectiveness and stealthiness poses a major limitation for existing attacks. To address this challenge, we proposeHidden Tail, a stealthy resource consumption attack that crafts prompt-agnostic adversarial images, inducing VLMs to generate maximum-length outputs by appending special tokens invisible to users. Our method employs a composite loss function that balances semantic preservation, repetitive special token induction, and suppression of the end-of-sequence (EOS) token, optimized via a dynamic weighting strategy. Extensive experiments show thatHidden Tailoutperforms existing attacks, increasing output length by up to 19.2× and reaching the maximum token limit, while preserving attack stealthiness. These results highlight the urgent need to improve the robustness of VLMs against efficiency-oriented adversarial threats. Our code is available athttps://github.com/zhangrui4041/Hidden_Tail.

An enormous volume of data is generated in the Internet of Things (IoT), which needs to be anonymized before sharing with public or third parties to minimize reidentification risk and protect sensitive information. Data anonymization techniques can remove information capable of identifying individuals. However, inappropriate data anonymization can increase the risk of reidentification. This work focuses on potential attack risks of anonymized data by evaluating the potential attack risks. Specifically, we analyzed the attack risks over anonymized data with both randomization and generalization techniques. We also analyzed the risk of reidentification for five commonly used data anonymization techniques. The experimental results demonstrate that the proposed solution can well evaluate the potential attack risks.