As large language models are deployed in high-stakes enterprise applications, from healthcare to finance, ensuring adherence to organization-specific policies has become essential. Yet existing safety evaluations focus exclusively on universal harms. We present COMPASS (Company/Organization Policy Alignment Assessment), the first systematic framework for evaluating whether LLMs comply with organizational allowlist and denylist policies. We apply COMPASS to eight diverse industry scenarios, generating and validating 5,920 queries that test both routine compliance and adversarial robustness through strategically designed edge cases. Evaluating seven state-of-the-art models, we uncover a fundamental asymmetry: models reliably handle legitimate requests (>95% accuracy) but catastrophically fail at enforcing prohibitions, refusing only 13–40% of adversarial denylist violations. These results demonstrate that current LLMs lack the robustness required for policy-critical deployments, establishing COMPASS as an essential evaluation framework for organizational AI safety.

Platooning is one of the promising technologies that maximizes the power efficiency of electric vehicles by decreasing the distances between the vehicles. Along with the development of autonomous driving technology, platooning is expected to be commercialized. Recent studies on the operation of platooning focused on power-efficient maintenance of platooning. However, power-efficient operation strategy is also needed for practical applications. Therefore, this study deals with platooning operations that can maximize the power efficiency of electric vehicles in various operational situations. In order to derive the operation method, a mathematical model structured with an objective function that minimizes power consumption is developed. To derive the solution of the mathematical model, a hybrid genetic algorithm is applied. The numerical experiments on four different operational situations are performed to verify the validity of the model and the solution procedure. The four situations consider overall situation that can happen during the platooning stage. The stages are formation, disassembly, join and breakaway of vehicles of platoon. Those four situations are decided upon since they can represent the general situation that can happen during platooning. As a result, the power-efficient driving patterns of electric vehicles are identified. After the development of electric and systematic technology, operational technology for platooning will collaborate for the further improvement. Therefore, throughout consideration of the formation of platooning, technology will expand the sustainability of technological development.

The news subheading summarizes an article’s contents in several sentences to support the headline limited to solely conveying the main contents. So, it is necessary to generate compelling news subheadings in consideration of the structural characteristics of the news. In this paper, we propose a subheading generation model using topical headline information. We introduce a discriminative learning method that utilizes the prediction result of masked headline tokens. Experiments show that the proposed model is effective and outperforms the comparative models on three news datasets written in two languages. We also show that our model performs robustly on a small dataset and various masking ratios. Qualitative analysis and human evaluations also show that the overall quality of generated subheadings improved over the comparative models.

Recent works have shown that attaching prompts to the input is effective at conditioning Language Models (LM) to perform specific tasks. However, prompts are always included in the input text during inference, even when they are fixed, thus incurring substantial computational and memory overhead. Also, there is currently no straightforward method of utilizing prompts that are longer than the maximum input length of the LMs without incurring additional costs during inference. We formally define Fixed Input Parameterization (FIP) problem that focuses on injecting the fixed prompt into the parameters of an LM to be an efficient alternative to attaching fixed prompts to the input. We show that in scenarios with long fixed prompts, FIP can be up to 280 times more efficient in terms of total FLOPs than previous approaches. We further explore methodologies for FIP and show promising results in persona-dependent conversation, semantic parsing, and zero-shot learning with task instructions. Through these explorations, we show that FIP can be a promising direction for conditioning language models, in scenarios with long and fixed prompts 1 .

News readers are left with their initial impression gained from its headline, whereas a headline's purpose isto attract the reader's attention to the news content. Incongruent news headlines can easily mislead readerswith their clickbait headlines, which have become pervasive in online. Thus, considerable attention has beenpaid to detect incongruent news headlines before they reach to the readers, however there is still a lack oflarge-scale dataset which is restricted to title and body text. Accordingly, in this study, we releasedIncongruentNewsHeadlineDataset, which has been collected and written by one of the largest news media outlets in SouthKorea. The generated dataset contains additional textual information, such as subtitles, image captions, andother auxiliary information. We proposed a method that effectively detects incongruent news headlines bycapturing the complex lexical and contextual textual relationships between a headline and its body using anattention mechanism. The proposed model outperforms the existing models, and we investigated how it fullyutilizes all the possible features.