Large Language Models (LLMs) suffer significant performance degradation in multi-turn conversations when information is presented incrementally. Given that multi-turn conversations characterize everyday interactions with LLMs, this degradation poses a severe challenge to real world usability. We hypothesize that abrupt increases in model uncertainty signal misalignment in multi-turn LLM interactions, and we exploit this insight to dynamically realign conversational context. We introduce ERGO (Entropy-guided Resetting for Generation Optimization), which continuously quantifies internal uncertainty via Shannon entropy over next token distributions and triggers adaptive prompt consolidation when a sharp spike in entropy is detected. By treating uncertainty as a first class signal rather than a nuisance to eliminate, ERGO embraces variability in language and modeling, representing and responding to uncertainty. In multi-turn tasks with incrementally revealed instructions, ERGO yields a 56.6% average performance gain over standard baselines, increases aptitude (peak performance capability) by 24.7%, and decreases unreliability (variability in performance) by 35.3%, demonstrating that uncertainty aware interventions can improve both accuracy and reliability in conversational AI.

The use of copyrighted books for training AI models has led to numerous lawsuits from authors concerned about AI's ability to generate derivative content. Yet it's unclear if these models can generate high quality literary text while emulating authors' styles. To answer this we conducted a preregistered study comparing MFA-trained expert writers with three frontier AI models: ChatGPT, Claude & Gemini in writing up to 450 word excerpts emulating 50 award-winning authors' diverse styles. In blind pairwise evaluations by 159 representative expert & lay readers, AI-generated text from in-context prompting was strongly disfavored by experts for both stylistic fidelity (OR=0.16, p<10^-8) & writing quality (OR=0.13, p<10^-7) but showed mixed results with lay readers. However, fine-tuning ChatGPT on individual authors' complete works completely reversed these findings: experts now favored AI-generated text for stylistic fidelity (OR=8.16, p<10^-13) & writing quality (OR=1.87, p=0.010), with lay readers showing similar shifts. These effects generalize across authors & styles. The fine-tuned outputs were rarely flagged as AI-generated (3% rate v. 97% for in-context prompting) by best AI detectors. Mediation analysis shows this reversal occurs because fine-tuning eliminates detectable AI stylistic quirks (e.g., cliche density) that penalize in-context outputs. While we do not account for additional costs of human effort required to transform raw AI output into cohesive, publishable prose, the median fine-tuning & inference cost of $81 per author represents a dramatic 99.7% reduction compared to typical professional writer compensation. Author-specific fine-tuning thus enables non-verbatim AI writing that readers prefer to expert human writing, providing empirical evidence directly relevant to copyright's fourth fair-use factor, the "effect upon the potential market or value" of the source works.

Large Language Models (LLMs) have demonstrated impressive reasoning capabilities but continue to struggle with arithmetic tasks. Prior works largely focus on outputs or prompting strategies, leaving the open question of the internal structure through which models do arithmetic computation. In this work, we investigate whether LLMs encode operator precedence in their internal representations via the open-source instruction-tuned LLaMA 3.2-3B model. We constructed a dataset of arithmetic expressions with three operands and two operators, varying the order and placement of parentheses. Using this dataset, we trace whether intermediate results appear in the residual stream of the instruction-tuned LLaMA 3.2-3B model. We apply interpretability techniques such as logit lens, linear classification probes, and UMAP geometric visualization. Our results show that intermediate computations are present in the residual stream, particularly after MLP blocks. We also find that the model linearly encodes precedence in each operator's embeddings post attention layer. We introduce partial embedding swap, a technique that modifies operator precedence by exchanging high-impact embedding dimensions between operators.

LLMs hold great promise for healthcare applications, but the rapid evolution of medical knowledge and errors in training data often cause them to generate outdated or inaccurate information, limiting their applicability in high-stakes clinical practice. Model editing has emerged as a potential remedy without full retraining. While parameter-based editing often compromises locality and is thus ill-suited for the medical domain, retrieval-based editing offers a more viable alternative. However, it still faces two critical challenges: (1) representation overlap within the medical knowledge space often causes inaccurate retrieval and reduces editing accuracy; (2) existing methods are restricted to single-sample edits, while batch-editing remains largely unexplored despite its importance for real-world medical applications. To address these challenges, we first construct MedVersa, an enhanced benchmark with broader coverage of medical subjects, designed to evaluate both single and batch edits under strict locality constraints. We then propose MedREK, a retrieval-based editing framework that integrates a shared query-key module for precise matching with an attention-based prompt encoder for informative guidance. Experimental results on various medical benchmarks demonstrate that our MedREK achieves superior performance across different core metrics and provides the first validated solution for batch-editing in medical LLMs. Our code and dataset are available at https://github.com/mylittleriver/MedREK.

The rapid spread of text generated by large language models (LLMs) makes it increasingly difficult to distinguish authentic human writing from machine output. Watermarking offers a promising solution: model owners can embed an imperceptible signal into generated text, marking its origin. Most leading approaches seed an LLM's next-token sampling with a pseudo-random key that can later be recovered to identify the text as machine-generated, while only minimally altering the model's output distribution. However, these methods suffer from two related issues: (i) watermarks are brittle to simple surface-level edits such as paraphrasing or reordering; and (ii) adversaries can append unrelated, potentially harmful text that inherits the watermark, risking reputational damage to model owners. To address these issues, we introduce SimKey, a semantic key module that strengthens watermark robustness by tying key generation to the meaning of prior context. SimKey uses locality-sensitive hashing over semantic embeddings to ensure that paraphrased text yields the same watermark key, while unrelated or semantically shifted text produces a different one. Integrated with state-of-the-art watermarking schemes, SimKey improves watermark robustness to paraphrasing and translation while preventing harmful content from false attribution, establishing semantic-aware keying as a practical and extensible watermarking direction.

The promise of LLM watermarking rests on a core assumption that a specific watermark proves authorship by a specific model. We demonstrate that this assumption is dangerously flawed. We introduce the threat of watermark spoofing, a sophisticated attack that allows a malicious model to generate text containing the authentic-looking watermark of a trusted, victim model. This enables the seamless misattribution of harmful content, such as disinformation, to reputable sources. The key to our attack is repurposing watermark radioactivity, the unintended inheritance of data patterns during fine-tuning, from a discoverable trait into an attack vector. By distilling knowledge from a watermarked teacher model, our framework allows an attacker to steal and replicate the watermarking signal of the victim model. This work reveals a critical security gap in text authorship verification and calls for a paradigm shift towards technologies capable of distinguishing authentic watermarks from expertly imitated ones. Our code is available at https://github.com/hsannn/ditto.git.

This paper proposes a modeling framework for dynamic topic evolution based on temporal large language models. The method first uses a large language model to obtain contextual embeddings of text and then introduces a temporal decay function and an attention mechanism. These components allow the model to adjust the importance of semantic units according to time intervals and capture topic variations across different periods. The temporal representations are then mapped into a latent topic space, where a state transition matrix is applied to describe the dynamic evolution of topics. A joint optimization objective constrains both semantic modeling and temporal consistency, ensuring diversity and smoothness in topic generation. The design emphasizes the unified modeling of semantic representation and temporal evolution, which improves topic coherence and diversity while enhancing stability and interpretability over time. Experiments on real-world corpora show that the framework effectively captures the generation, expansion, and decline of topics and outperforms existing models across multiple metrics. Overall, the proposed method provides a systematic solution for understanding dynamic semantic patterns in large-scale text, enriches the research paradigm of topic modeling, and supports complex text analysis tasks in multiple domains.

Large Language Models are increasingly adopted in financial applications to support investment workflows. However, prior studies have seldom examined how these models reflect biases related to firm size, sector, or financial characteristics, which can significantly impact decision-making. This paper addresses this gap by focusing on representation bias in open-source Qwen models. We propose a balanced round-robin prompting method over approximately 150 U.S. equities, applying constrained decoding and token-logit aggregation to derive firm-level confidence scores across financial contexts. Using statistical tests and variance analysis, we find that firm size and valuation consistently increase model confidence, while risk factors tend to decrease it. Confidence varies significantly across sectors, with the Technology sector showing the greatest variability. When models are prompted for specific financial categories, their confidence rankings best align with fundamental data, moderately with technical signals, and least with growth indicators. These results highlight representation bias in Qwen models and motivate sector-aware calibration and category-conditioned evaluation protocols for safe and fair financial LLM deployment.

Language model finetuning often results in learning undesirable traits in combination with desired ones. To address this, we propose inoculation prompting: modifying finetuning data by prepending a short system-prompt instruction that deliberately elicits the undesirable trait. At test time, we evaluate without the instruction; inoculated models have much lower expression of the trait than models trained with unmodified training data. Inoculation is selective: in a toy setting where assistant responses are always in Spanish and ALL-CAPS, an appropriate inoculation (e.g., ``You always speak in Spanish.'') teaches the model to capitalize responses while still responding in English. We find that inoculation is also effective across several additional settings: reducing emergent misalignment (EM) from task-specific finetuning, defending against backdoor injections, and mitigating the transmission of traits via subliminal learning. Follow-up analysis suggests a mechanism: making a trait less surprising via inoculation reduces optimization pressure to globally update the model, thereby reducing the degree of generalization. Our analysis relates to prior work on EM: inoculation explains prior findings that educational contexts mitigate EM from insecure code. Beyond demonstrating a simple and effective technique for selective learning, our results contribute to a better conceptual understanding of how and why language models generalize.

Amidst the rapid normalization of generative artificial intelligence (GAI), intelligent systems have come to dominate political discourse across information media. However, internalized political biases stemming from training data skews, human prejudice, and algorithmic flaws continue to plague this novel technology. This study employs a zero-shot classification approach to evaluate algorithmic political partisanship through a methodical combination of ideological alignment, topicality, response sentiment, and objectivity. A total of 1800 model responses across six mainstream large language models (LLMs) were individually input into four distinct fine-tuned classification algorithms, each responsible for computing one of the aforementioned metrics. The results show an amplified liberal-authoritarian alignment across the six LLMs evaluated, with notable instances of reasoning supersessions and canned refusals. The study subsequently highlights the psychological influences underpinning human-computer interactions and how intrinsic biases can permeate public discourse. The resulting distortion of the political landscape can ultimately manifest as conformity or polarization, depending on the region's pre-existing socio-political structures.