Large language models are increasingly used for creative writing and engagement content, raising safety concerns about their outputs. Using humor generation as a testbed, this work evaluates how funniness optimization in modern LLM pipelines couples with harmful content by jointly measuring humor, stereotypicality, and toxicity. We further supplement this by analyzing incongruity signals through information-theoretic metrics. Across six models, we observe that even for fixed neutral setups, harmful outputs receive higher humor scores, indicating a bias amplification loop between generators and evaluators. Information-theoretic analyses show that harmful cues widen predictive uncertainty and, surprisingly, can even make harmful punchlines more expected for some models, suggesting intrinsic structural embedding in learned humor distributions. Experiments and human evaluation on an additional satire-generation task with human-perceived funniness judgments show that LLM funniness relies on increased stereotypicality and toxicity, including for closed models. Quantitatively, stereotypical/toxic jokes gain 10%–21% in mean humor score, stereotypical jokes appear 11% to 28% more often among the jokes marked funny by an LLM-based metric, and up to 10% more often in generations perceived as funny by humans.

We explore the ability of large language models (LLMs) to engage in subtle deception through strategically phrasing and intentionally manipulating information. This harmful behavior can be hard to detect, unlike blatant lying or unintentional hallucination. We build a simple testbed mimicking a legislative environment where a corporate lobbyist module is proposing amendments to bills that benefit a specific company while evading identification of this benefactor. We use real-world legislative bills matched with potentially affected companies to ground these interactions. Our results show that LLM lobbyists can draft subtle phrasing to avoid such identification by strong LLM-based detectors. Further optimization of the phrasing using LLM-based re-planning and re-sampling increases deception rates by up to 40 percentage points.Our human evaluations to verify the quality of deceptive generations and their retention of self-serving intent show significant coherence with our automated metrics and also help in identifying certain strategies of deceptive phrasing.This study highlights the risk of LLMs’ capabilities for strategic phrasing through seemingly neutral language to attain self-serving goals. This calls for future research to uncover and protect against such subtle deception.

Named Entity Recognition (NER), an essential subtask in NLP that identifies text belonging to predefined semantics such as a person, location, organization, drug, time, clinical procedure, biological protein, etc. NER plays a vital role in various fields such as informationextraction, question answering, and machine translation. This paper describes our participating system run to the Named entity recognitionand classification shared task SemEval-2022. The task is motivated towards detecting semantically ambiguous and complex entities in shortand low-context settings. Our team focused on improving entity recognition by improving the word embeddings. We concatenated the word representations from State-of-the-art language models and passed them to find the best representation through a reinforcement trainer. Our results highlight the improvements achieved by various embedding concatenations.