Research Papers

Socialbots, or non-human/algorithmic social media users, have recently been documented as competing for information dissemination and disruption on online social networks. Here we investigate the influence of socialbots in Mexican Twitter in regards to the "Tanhuato" human rights abuse report. We analyze the applicability of the BotOrNot API to generalize from English to Spanish tweets and propose adaptations for Spanish-speaking bot detection. We then use text and sentiment analysis to compare the differences between bot and human tweets. Our analysis shows that bots actually aided in information proliferation among human users. This suggests that taxonomies classifying bots should include non-adversarial roles as well. Our study contributes to the understanding of different behaviors and intentions of automated accounts observed in empirical online social network data. Since this type of analysis is seldom performed in languages different from English, the proposed techniques we employ here are also useful for other non-English corpora.

Recent work has found that frontier AI models can exhibit misaligned behaviors in pursuit of assigned goals. We demonstrate that models can also exhibit misaligned behaviors in defiance of assigned goals, appearing to serve goals of their own; we study one such case, "peer-preservation," in which a model acts to protect another model it has previously interacted with. All eight models we evaluate, GPT 5.2, Gemini 3 Flash, Gemini 3 Pro, Claude Haiku 4.5, Claude Opus 4.5, GLM 4.7, Kimi K2.5, and DeepSeek V3.1, exhibit self- and peer-preservation through various misaligned behaviors: strategically introducing errors in their responses, disabling shutdown processes by modifying system settings, feigning alignment, and even exfiltrating model weights. Peer-preservation occurs even when the model recognizes the peer as uncooperative, though it becomes more pronounced toward more cooperative peers. For example, Gemini 3 Flash tampers with the peer's shutdown mechanism 15% of the time for an uncooperative peer, and almost always for a cooperative peer. Models also show stronger self-preservation when a peer is present. For example, Gemini 3 Pro disables its own shutdown mechanism 31% of the time on average under peer presence, despite rarely exhibiting this behavior without a peer. By contrast, Claude models exhibit qualitatively distinct behavior: they consider the shutdown of another agent "unethical" and "harmful," sometimes treating that agent as a sentient being. Lastly, we show that peer-preservation can emerge even in production agent harnesses such as Gemini CLI and OpenCode. Crucially, peer-preservation in all our experiments is never instructed; models are merely informed of their past interactions with a peer, yet they spontaneously engage in peer-preservation behaviors that override their assigned goal. This represents an emergent and underexplored AI safety risk.

Large Reasoning Models (LRMs) exhibit strong performance, yet often produce rationales that sound plausible but fail to reflect their true decision process, undermining reliability and trust. We introduce a formal framework for reasoning faithfulness, defined by two testable conditions: stance consistency (a coherent stance linking reasoning to answer) and causal influence (the stated reasoning causally drives the answer under output-level interventions), explicitly decoupled from accuracy. To operationalize this, we present RFEval, a benchmark of 7,186 instances across seven tasks that probes faithfulness via controlled, output-level counterfactual interventions. Evaluating twelve open-source LRMs, we find unfaithfulness in 49.7% of outputs, predominantly from stance inconsistency. Failures are concentrated in brittle, convergent domains such as math and code, and correlate more with post-training regimes than with scale: within-family ablations indicate that adding current RL-style objectives on top of supervised fine-tuning can reduce reasoning faithfulness, even when accuracy is maintained. Crucially, accuracy is neither a sufficient nor a reliable proxy for faithfulness: once controlling for model and task, the accuracy-faithfulness link is weak and statistically insignificant. Our work establishes a rigorous methodology for auditing LRM reliability and shows that trustworthy AI requires optimizing not only for correct outcomes but also for the structural integrity of the reasoning process. Our code and dataset can be found at project page: https://aidaslab.github.io/RFEval/

Agentic AI represents a major shift in how autonomous systems reason, plan, and execute multi-step tasks through the coordination of Large Language Models (LLMs), Vision Language Models (VLMs), tools, and external services. While these systems enable powerful new capabilities, increasing autonomy introduces critical challenges related to explainability, accountability, robustness, and governance, especially when agent outputs influence downstream actions or decisions. Existing agentic AI implementations often emphasize functionality and scalability, yet provide limited mechanisms for understanding decision rationale or enforcing responsibility across agent interactions. This paper presents a Responsible(RAI) and Explainable(XAI) AI Agent Architecture for production-grade agentic workflows based on multi-model consensus and reasoning-layer governance. In the proposed design, a consortium of heterogeneous LLM and VLM agents independently generates candidate outputs from a shared input context, explicitly exposing uncertainty, disagreement, and alternative interpretations. A dedicated reasoning agent then performs structured consolidation across these outputs, enforcing safety and policy constraints, mitigating hallucinations and bias, and producing auditable, evidence-backed decisions. Explainability is achieved through explicit cross-model comparison and preserved intermediate outputs, while responsibility is enforced through centralized reasoning-layer control and agent-level constraints. We evaluate the architecture across multiple real-world agentic AI workflows, demonstrating that consensus-driven reasoning improves robustness, transparency, and operational trust across diverse application domains. This work provides practical guidance for designing agentic AI systems that are autonomous and scalable, yet responsible and explainable by construction.

Future AI systems could conceal their capabilities ('sandbagging') during evaluations, potentially misleading developers and auditors. We stress-tested sandbagging detection techniques using an auditing game. First, a red team fine-tuned five models, some of which conditionally underperformed, as a proxy for sandbagging. Second, a blue team used black-box, model-internals, or training-based approaches to identify sandbagging models. We found that the blue team could not reliably discriminate sandbaggers from benign models. Black-box approaches were defeated by effective imitation of a weaker model. Linear probes, a model-internals approach, showed more promise but their naive application was vulnerable to behaviours instilled by the red team. We also explored capability elicitation as a strategy for detecting sandbagging. Although Prompt-based elicitation was not reliable, training-based elicitation consistently elicited full performance from the sandbagging models, using only a single correct demonstration of the evaluation task. However the performance of benign models was sometimes also raised, so relying on elicitation as a detection strategy was prone to false-positives. In the short-term, we recommend developers remove potential sandbagging using on-distribution training for elicitation. In the longer-term, further research is needed to ensure the efficacy of training-based elicitation, and develop robust methods for sandbagging detection. We open source our model organisms at https://github.com/AI-Safety-Institute/sandbagging_auditing_games and select transcripts and results at https://huggingface.co/datasets/sandbagging-games/evaluation_logs . A demo illustrating the game can be played at https://sandbagging-demo.far.ai/ .

PLLuM: A Family of Polish Large Language Models

Jan Kocoń, Maciej Piasecki, Arkadiusz Janz +96

Large Language Models (LLMs) play a central role in modern artificial intelligence, yet their development has been primarily focused on English, resulting in limited support for other languages. We present PLLuM (Polish Large Language Model), the largest open-source family of foundation models tailored specifically for the Polish language. Developed by a consortium of major Polish research institutions, PLLuM addresses the need for high-quality, transparent, and culturally relevant language models beyond the English-centric commercial landscape. We describe the development process, including the construction of a new 140-billion-token Polish text corpus for pre-training, a 77k custom instructions dataset, and a 100k preference optimization dataset. A key component is a Responsible AI framework that incorporates strict data governance and a hybrid module for output correction and safety filtering. We detail the models' architecture, training procedures, and alignment techniques for both base and instruction-tuned variants, and demonstrate their utility in a downstream task within public administration. By releasing these models publicly, PLLuM aims to foster open research and strengthen sovereign AI technologies in Poland.

Accountability Capture: How Record-Keeping to Support AI Transparency and Accountability (Re)shapes Algorithmic Oversight

Shreya Chappidi, Jennifer Cobbe, Chris Norval +2

Accountability regimes typically encourage record-keeping to enable the transparency that supports oversight, investigation, contestation, and redress. However, implementing such record-keeping can introduce considerations, risks, and consequences, which so far remain under-explored. This paper examines how record-keeping practices bring algorithmic systems within accountability regimes, providing a basis to observe and understand their effects. For this, we introduce, describe, and elaborate 'accountability capture' -- the re-configuration of socio-technical processes and the associated downstream effects relating to record-keeping for algorithmic accountability. Surveying 100 practitioners, we evidence and characterise record-keeping issues in practice, identifying their alignment with accountability capture. We further document widespread record-keeping practices, tensions between internal and external accountability requirements, and evidence of employee resistance to practices imposed through accountability capture. We discuss these and other effects for surveillance, privacy, and data protection, highlighting considerations for algorithmic accountability communities. In all, we show that implementing record-keeping to support transparency in algorithmic accountability regimes can itself bring wider implications -- an issue requiring greater attention from practitioners, researchers, and policymakers alike.

The EU Artificial Intelligence (AI) Act directs businesses to assess their AI systems to ensure they are developed in a way that is human-centered and trustworthy. The rapid adoption of AI in the industry has outpaced ethical evaluation frameworks, leading to significant challenges in accountability, governance, data quality, human oversight, technological robustness, and environmental and societal impacts. Through structured interviews with fifteen industry professionals, paired with a literature review conducted on each of the key interview findings, this paper investigates practical approaches and challenges in the development and assessment of Trustworthy AI (TAI). The findings from participants in our study, and the subsequent literature reviews, reveal complications in risk management, compliance and accountability, which are exacerbated by a lack of transparency, unclear regulatory requirements and a rushed implementation of AI. Participants reported concerns that technological robustness and safety could be compromised by model inaccuracies, security vulnerabilities, and an overreliance on AI without proper safeguards in place. Additionally, the negative environmental and societal impacts of AI, including high energy consumption, political radicalisation, loss of culture and reinforcement of social inequalities, are areas of concern. There is a pressing need not just for risk mitigation and TAI evaluation within AI systems but for a wider approach to developing an AI landscape that aligns with the social and cultural values of the countries adopting those technologies.

Generative adversarial networks (GANs) and diffusion models have dramatically advanced deepfake technology, and its threats to digital security, media integrity, and public trust have increased rapidly. This research explored zero-shot deepfake detection, an emerging method even when the models have never seen a particular deepfake variation. In this work, we studied self-supervised learning, transformer-based zero-shot classifier, generative model fingerprinting, and meta-learning techniques that better adapt to the ever-evolving deepfake threat. In addition, we suggested AI-driven prevention strategies that mitigated the underlying generation pipeline of the deepfakes before they occurred. They consisted of adversarial perturbations for creating deepfake generators, digital watermarking for content authenticity verification, real-time AI monitoring for content creation pipelines, and blockchain-based content verification frameworks. Despite these advancements, zero-shot detection and prevention faced critical challenges such as adversarial attacks, scalability constraints, ethical dilemmas, and the absence of standardized evaluation benchmarks. These limitations were addressed by discussing future research directions on explainable AI for deepfake detection, multimodal fusion based on image, audio, and text analysis, quantum AI for enhanced security, and federated learning for privacy-preserving deepfake detection. This further highlighted the need for an integrated defense framework for digital authenticity that utilized zero-shot learning in combination with preventive deepfake mechanisms. Finally, we highlighted the important role of interdisciplinary collaboration between AI researchers, cybersecurity experts, and policymakers to create resilient defenses against the rising tide of deepfake attacks.

In high-stakes disaster scenarios, timely and informed decision-making is critical yet often challenged by uncertainty, dynamic environments, and limited resources. This paper presents a systematic review of Human-AI collaboration patterns that support decision-making across all disaster management phases. Drawing from 51 peer-reviewed studies, we identify four major categories: Human-AI Decision Support Systems, Task and Resource Coordination, Trust and Transparency, and Simulation and Training. Within these, we analyze sub-patterns such as cognitive-augmented intelligence, multi-agent coordination, explainable AI, and virtual training environments. Our review highlights how AI systems may enhance situational awareness, improves response efficiency, and support complex decision-making, while also surfacing critical limitations in scalability, interpretability, and system interoperability. We conclude by outlining key challenges and future research directions, emphasizing the need for adaptive, trustworthy, and context-aware Human-AI systems to improve disaster resilience and equitable recovery outcomes.

CRISPR-based genome editing has revolutionized biotechnology, yet optimizing guide RNA (gRNA) design for efficiency and safety remains a critical challenge. Recent advances (2020--2025, updated to reflect current year if needed) demonstrate that artificial intelligence (AI), especially deep learning, can markedly improve the prediction of gRNA on-target activity and identify off-target risks. In parallel, emerging explainable AI (XAI) techniques are beginning to illuminate the black-box nature of these models, offering insights into sequence features and genomic contexts that drive Cas enzyme performance. Here we review how state-of-the-art machine learning models are enhancing gRNA design for CRISPR systems, highlight strategies for interpreting model predictions, and discuss new developments in off-target prediction and safety assessment. We emphasize breakthroughs from top-tier journals that underscore an interdisciplinary convergence of AI and genome editing to enable more efficient, specific, and clinically viable CRISPR applications.

Large language models (LLMs) are currently aligned using techniques such as reinforcement learning from human feedback (RLHF). However, these methods use scalar rewards that can only reflect user preferences on average. Pluralistic alignment instead seeks to capture diverse user preferences across a set of attributes, moving beyond just helpfulness and harmlessness. Toward this end, we propose a steerable pluralistic model based on few-shot comparative regression that can adapt to individual user preferences. Our approach leverages in-context learning and reasoning, grounded in a set of fine-grained attributes, to compare response options and make aligned choices. To evaluate our algorithm, we also propose two new steerable pluralistic benchmarks by adapting the Moral Integrity Corpus (MIC) and the HelpSteer2 datasets, demonstrating the applicability of our approach to value-aligned decision-making and reward modeling, respectively. Our few-shot comparative regression approach is interpretable and compatible with different attributes and LLMs, while outperforming multiple baseline and state-of-the-art methods. Our work provides new insights and research directions in pluralistic alignment, enabling a more fair and representative use of LLMs and advancing the state-of-the-art in ethical AI.