Current safety alignment for large language models(LLMs) continues to present vulnerabilities, given that adversarial prompting can effectively bypass their safety measures.Our investigation shows that these safety mechanisms predominantly depend on a limited subset of attention heads: removing or ablating these heads can severely compromise model safety. To identify and evaluate these safety-critical components, we introduce RDSHA, a targeted ablation method that leverages the model's refusal direction to pinpoint attention heads mostly responsible for safety behaviors. Further analysis shows that existing jailbreak attacks exploit this concentration by selectively bypassing or manipulating these critical attention heads. To address this issue, we propose AHD, a novel training strategy designed to promote the distributed encoding of safety-related behaviors across numerous attention heads. Experimental results demonstrate that AHD successfully distributes safety-related capabilities across more attention heads. Moreover, evaluations under several mainstream jailbreak attacks show that models trained with AHD exhibit considerably stronger safety robustness, while maintaining overall functional utility.

This thesis examines the obstacles and potential solutions for creating such bots, focusing on innovative techniques that enable bots to learn and adapt autonomously in constantly changing environments. End-to-end task bots, typically built using a static and limited corpus, face difficulties when deployed online due to three primary factors tied to this limitation. First, they might confront queries featuring unexpected linguistic patterns or slot values (i.e., unseen user behaviors). Second, they could potentially face requirements for new functions or tasks (i.e., task definition extensions). Third, even when equipped with relevant knowledge, these bots may produce responses that appear plausible but are actually incorrect (i.e., "hallucinations"). Addressing these challenges is vital for enhancing task bots' performance and reliability in real-world settings.

Graph-structured data, which captures non-Euclidean relationships and interactions between entities, is growing in scale and complexity. As a result, training state-of-the-art graph machine learning (GML) models have become increasingly resource-intensive, turning these models and data into invaluable Intellectual Property (IP). To address the resource-intensive nature of model training, graph-based Machine-Learning-as-a-Service (GMLaaS) has emerged as an efficient solution by leveraging third-party cloud services for model development and management. However, deploying such models in GMLaaS also exposes them to potential threats from attackers. Specifically, while the APIs within a GMLaaS system provide interfaces for users to query the model and receive outputs, they also allow attackers to exploit and steal model functionalities or sensitive training data, posing severe threats to the safety of these GML models and the underlying graph data. To address these challenges, this survey systematically introduces the first taxonomy of threats and defenses at the level of both GML model and graph-structured data. Such a tailored taxonomy facilitates an in-depth understanding of GML IP protection. Furthermore, we present a systematic evaluation framework to assess the effectiveness of IP protection methods, introduce a curated set of benchmark datasets across various domains, and discuss their application scopes and future challenges. Finally, we establish an open-sourced versatile library named PyGIP, which evaluates various attack and defense techniques in GMLaaS scenarios and facilitates the implementation of existing benchmark methods. The library resource can be accessed at: https://labrai.github.io/PyGIP. We believe this survey will play a fundamental role in intellectual property protection for GML and provide practical recipes for the GML community.

Rapid LLM advancements heighten fake news risks by enabling the automatic generation of increasingly sophisticated misinformation. Previous detection methods, including fine-tuned small models or LLM-based detectors, often struggle with its dynamically evolving nature. In this work, we propose a novel framework called the Symbolic Adversarial Learning Framework (SALF), which implements an adversarial training paradigm by an agent symbolic learning optimization process, rather than relying on numerical updates. SALF introduces a paradigm where the generation agent crafts deceptive narratives, and the detection agent uses structured debates to identify logical and factual flaws for detection, and they iteratively refine themselves through such adversarial interactions. Unlike traditional neural updates, we represent agents using agent symbolic learning, where learnable weights are defined by agent prompts, and simulate back-propagation and gradient descent by operating on natural language representations of weights, loss, and gradients. Experiments on two multilingual benchmark datasets demonstrate SALF's effectiveness, showing it generates sophisticated fake news that degrades state-of-the-art detection performance by up to 53.4% in Chinese and 34.2% in English on average. SALF also refines detectors, improving detection of refined content by up to 7.7%. We hope our work inspires further exploration into more robust, adaptable fake news detection systems.

Generative artificial intelligence in music has made significant strides, yet it still falls short of the substantial achievements seen in natural language processing, primarily due to the limited availability of music data. Knowledge-informed approaches have been shown to enhance the performance of music generation models, even when only a few pieces of musical knowledge are integrated. This paper seeks to leverage comprehensive music theory in AI-driven music generation tasks, such as algorithmic composition and style transfer, which traditionally require significant manual effort with existing techniques. We introduce a novel automatic music lexicon construction model that generates a lexicon, named CompLex, comprising 37,432 items derived from just 9 manually input category keywords and 5 sentence prompt templates. A new multi-agent algorithm is proposed to automatically detect and mitigate hallucinations. CompLex demonstrates impressive performance improvements across three state-of-the-art text-to-music generation models, encompassing both symbolic and audio-based methods. Furthermore, we evaluate CompLex in terms of completeness, accuracy, non-redundancy, and executability, confirming that it possesses the key characteristics of an effective lexicon.

This paper identifies and analyzes a novel vulnerability class in Model Context Protocol (MCP) based agent systems. The attack chain describes and demonstrates how benign, individually authorized tasks can be orchestrated to produce harmful emergent behaviors. Through systematic analysis using the MITRE ATLAS framework, we demonstrate how 95 agents tested with access to multiple services-including browser automation, financial analysis, location tracking, and code deployment-can chain legitimate operations into sophisticated attack sequences that extend beyond the security boundaries of any individual service. These red team exercises survey whether current MCP architectures lack cross-domain security measures necessary to detect or prevent a large category of compositional attacks. We present empirical evidence of specific attack chains that achieve targeted harm through service orchestration, including data exfiltration, financial manipulation, and infrastructure compromise. These findings reveal that the fundamental security assumption of service isolation fails when agents can coordinate actions across multiple domains, creating an exponential attack surface that grows with each additional capability. This research provides a barebones experimental framework that evaluate not whether agents can complete MCP benchmark tasks, but what happens when they complete them too well and optimize across multiple services in ways that violate human expectations and safety constraints. We propose three concrete experimental directions using the existing MCP benchmark suite.

Objectivity in journalism has long been contested, oscillating between ideals of neutral, fact-based reporting and the inevitability of subjective framing. With the advent of large language models (LLMs), these tensions are now mediated by algorithmic systems whose training data and design choices may themselves embed cultural or ideological biases. This study investigates geopolitical parallax-systematic divergence in news quality and subjectivity assessments-by comparing article-level embeddings from Chinese-origin (Qwen, BGE, Jina) and Western-origin (Snowflake, Granite) model families. We evaluate both on a human-annotated news quality benchmark spanning fifteen stylistic, informational, and affective dimensions, and on parallel corpora covering politically sensitive topics, including Palestine and reciprocal China-United States coverage. Using logistic regression probes and matched-topic evaluation, we quantify per-metric differences in predicted positive-class probabilities between model families. Our findings reveal consistent, non-random divergences aligned with model origin. In Palestine-related coverage, Western models assign higher subjectivity and positive emotion scores, while Chinese models emphasize novelty and descriptiveness. Cross-topic analysis shows asymmetries in structural quality metrics Chinese-on-US scoring notably lower in fluency, conciseness, technicality, and overall quality-contrasted by higher negative emotion scores. These patterns align with media bias theory and our distinction between semantic, emotional, and relational subjectivity, and extend LLM bias literature by showing that geopolitical framing effects persist in downstream quality assessment tasks. We conclude that LLM-based media evaluation pipelines require cultural calibration to avoid conflating content differences with model-induced bias.

We propose a transformer-based rhythm quantization model that incorporates beat and downbeat information to quantize MIDI performances into metrically-aligned, human-readable scores. We propose a beat-based preprocessing method that transfers score and performance data into a unified token representation. We optimize our model architecture and data representation and train on piano and guitar performances. Our model exceeds state-of-the-art performance based on the MUSTER metric.

Symbolic music generation has seen rapid progress with artificial neural networks, yet remains underexplored in the biologically plausible domain of spiking neural networks (SNNs), where both standardized benchmarks and comprehensive evaluation methods are lacking. To address this gap, we introduce MuSpike, a unified benchmark and evaluation framework that systematically assesses five representative SNN architectures (SNN-CNN, SNN-RNN, SNN-LSTM, SNN-GAN and SNN-Transformer) across five typical datasets, covering tonal, structural, emotional, and stylistic variations. MuSpike emphasizes comprehensive evaluation, combining established objective metrics with a large-scale listening study. We propose new subjective metrics, targeting musical impression, autobiographical association, and personal preference, that capture perceptual dimensions often overlooked in prior work. Results reveal that (1) different SNN models exhibit distinct strengths across evaluation dimensions; (2) participants with different musical backgrounds exhibit diverse perceptual patterns, with experts showing greater tolerance toward AI-composed music; and (3) a noticeable misalignment exists between objective and subjective evaluations, highlighting the limitations of purely statistical metrics and underscoring the value of human perceptual judgment in assessing musical quality. MuSpike provides the first systematic benchmark and systemic evaluation framework for SNN models in symbolic music generation, establishing a solid foundation for future research into biologically plausible and cognitively grounded music generation.

As generative AI continues to reshape artistic production and alternate modes of human expression, artists whose livelihoods are most directly affected have raised urgent concerns about consent, transparency, and the future of creative labor. However, the voices of artists are often marginalized in dominant public and scholarly discourse. This study presents a twelve-year analysis, from 2013 to 2025, of English-language discourse surrounding AI-generated art. It draws from 439 curated 500-word excerpts sampled from opinion articles, news reports, blogs, legal filings, and spoken-word transcripts. Through a reproducible methodology, we identify five stable thematic clusters and uncover a misalignment between artists' perceptions and prevailing media narratives. Our findings highlight how the use of technical jargon can function as a subtle form of gatekeeping, often sidelining the very issues artists deem most urgent. Our work provides a BERTopic-based methodology and a multimodal baseline for future research, alongside a clear call for deeper, transparency-driven engagement with artist perspectives in the evolving AI-creative landscape.