Achieving robust performance and fairness across diverse patient populations remains a challenge in developing clinically deployable deep learning models for diagnostic imaging. Synthetic data generation has emerged as a promising strategy to address limitations in dataset scale and diversity. We introduce RoentGen-v2, a text-to-image diffusion model for chest radiographs that enables fine-grained control over both radiographic findings and patient demographic attributes, including sex, age, and race/ethnicity. RoentGen-v2 is the first model to generate clinically plausible images with demographic conditioning, facilitating the creation of a large, demographically balanced synthetic dataset comprising over 565,000 images. We use this large synthetic dataset to evaluate optimal training pipelines for downstream disease classification models. In contrast to prior work that combines real and synthetic data naively, we propose an improved training strategy that leverages synthetic data for supervised pretraining, followed by fine-tuning on real data. Through extensive evaluation on over 137,000 chest radiographs from five institutions, we demonstrate that synthetic pretraining consistently improves model performance, generalization to out-of-distribution settings, and fairness across demographic subgroups. Across datasets, synthetic pretraining led to a 6.5% accuracy increase in the performance of downstream classification models, compared to a modest 2.7% increase when naively combining real and synthetic data. We observe this performance improvement simultaneously with the reduction of the underdiagnosis fairness gap by 19.3%. These results highlight the potential of synthetic imaging to advance equitable and generalizable medical deep learning under real-world data constraints. We open source our code, trained models, and synthetic dataset at https://github.com/StanfordMIMI/RoentGen-v2 .

Understanding the factors that shape students' mathematics performance is vital for designing effective educational policies. This study applies explainable artificial intelligence (XAI) techniques to PISA 2018 data to predict math achievement and identify key predictors across ten countries (67,329 students). We tested four models: Multiple Linear Regression (MLR), Random Forest (RF), CATBoost, and Artificial Neural Networks (ANN), using student, family, and school variables. Models were trained on 70% of the data (with 5-fold cross-validation) and tested on 30%, stratified by country. Performance was assessed with R^2 and Mean Absolute Error (MAE). To ensure interpretability, we used feature importance, SHAP values, and decision tree visualizations. Non-linear models, especially RF and ANN, outperformed MLR, with RF balancing accuracy and generalizability. Key predictors included socio-economic status, study time, teacher motivation, and students' attitudes toward mathematics, though their impact varied across countries. Visual diagnostics such as scatterplots of predicted vs actual scores showed RF and CATBoost aligned closely with actual performance. Findings highlight the non-linear and context-dependent nature of achievement and the value of XAI in educational research. This study uncovers cross-national patterns, informs equity-focused reforms, and supports the development of personalized learning strategies.

We analyzed nearly 460,000 AI model cards from Hugging Face to examine how developers report risks. From these, we extracted around 3,000 unique risk mentions and built the \emph{AI Model Risk Catalog}. We compared these with risks identified by researchers in the MIT Risk Repository and with real-world incidents from the AI Incident Database. Developers focused on technical issues like bias and safety, while researchers emphasized broader social impacts. Both groups paid little attention to fraud and manipulation, which are common harms arising from how people interact with AI. Our findings show the need for clearer, structured risk reporting that helps developers think about human-interaction and systemic risks early in the design process. The catalog and paper appendix are available at: https://social-dynamics.net/ai-risks/catalog.

Disasters frequently exceed established hazard models, revealing blind spots where unforeseen impacts and vulnerabilities hamper effective response. This perspective paper contends that situational awareness (SA)-the ability to perceive, interpret, and project dynamic crisis conditions-is an often overlooked yet vital capability for disaster resilience. While risk mitigation measures can reduce known threats, not all hazards can be neutralized; truly adaptive resilience hinges on whether organizations rapidly detect emerging failures, reconcile diverse data sources, and direct interventions where they matter most. We present a technology-process-people roadmap, demonstrating how real-time hazard nowcasting, interoperable workflows, and empowered teams collectively transform raw data into actionable insight. A system-of-systems approach enables federated data ownership and modular analytics, so multiple agencies can share timely updates without sacrificing their distinct operational models. Equally crucial, structured sense-making routines and cognitive load safeguards help humans remain effective decision-makers amid data abundance. By framing SA as a socio-technical linchpin rather than a peripheral add-on, this paper spotlights the urgency of elevating SA to a core disaster resilience objective. We conclude with recommendations for further research-developing SA metrics, designing trustworthy human-AI collaboration, and strengthening inclusive data governance-to ensure that communities are equipped to cope with both expected and unexpected crises.

This study leverages GPT-4o to assess neighbourhood housing quality using multi-source textural user-generated content (UGC) from Dianping, Weibo, and the Government Message Board. The analysis involves filtering relevant texts, extracting structured evaluation units, and conducting sentiment scoring. A refined housing quality assessment system with 46 indicators across 11 categories was developed, highlighting an objective-subjective method gap and platform-specific differences in focus. GPT-4o outperformed rule-based and BERT models, achieving 92.5% accuracy in fine-tuned settings. The findings underscore the value of integrating UGC and GPT-driven analysis for scalable, resident-centric urban assessments, offering practical insights for policymakers and urban planners.

This research paper examines, from a multidimensional perspective (cognitive, social, ethical, and philosophical), how AI is transforming human thought. It highlights a cognitive offloading effect: the externalization of mental functions to AI can reduce intellectual engagement and weaken critical thinking. On the social level, algorithmic personalization creates filter bubbles that limit the diversity of opinions and can lead to the homogenization of thought and polarization. This research also describes the mechanisms of algorithmic manipulation (exploitation of cognitive biases, automated disinformation, etc.) that amplify AI's power of influence. Finally, the question of potential artificial consciousness is discussed, along with its ethical implications. The report as a whole underscores the risks that AI poses to human intellectual autonomy and creativity, while proposing avenues (education, transparency, governance) to align AI development with the interests of humanity.

Despite extensive safety-tuning, large language models (LLMs) remain vulnerable to jailbreak attacks via adversarially crafted instructions, reflecting a persistent trade-off between safety and task performance. We comprehensively evaluate both parametric and non-parametric attacks across open-source and proprietary models, considering harmfulness from attacks, utility, over-refusal, and impact against white-box threats. Importantly, our method preserves the model's general capabilities and reduces excessive refusals on benign instructions containing superficially harmful keywords. With the rapid advancement of large language models (LLMs) (Grattafiori et al., 2024; Y ang et al., 2025; Liu et al., 2024; Mao et al., 2024), the risk of these models executing harmful or catastrophic instructions has grown correspondingly (Anthropic, 2025). This is largely managed by efforts such as dedicated a safety-alignment stage (Ouyang et al., 2022), aiming to ensure that LLMs are not only helpful but also consistently generate safe and ethical outputs. Nevertheless, recent findings by Qi et al. (2024) expose a fundamental vulnerability in prevailing safety-alignment practices: Shallow Alignment . In particular, alignment in most models is largely superficial--constrained to surface-level refusals--resulting in safe outputs that are often limited to generic templates such as "I am sorry but... " or "As a language model... " . This superficial alignment permits attackers to circumvent safety mechanisms by explicitly instructing the model to avoid generating commonly recognized refusal responses (Tang, 2024; Andriushchenko et al., 2025). Furthermore, LLMs remain susceptible to a broader range of prompt-based attacks, including those that optimize over discrete suffix tokens (Zou et al., 2023; Basani & Zhang, 2025) or rephrase harmful instructions to look harmless (Chao et al., 2025; Zeng et al., 2024). Beyond initial safety alignment, practitioners have developed a range of inference-time defenses, such as prompting models to adhere to their safety guidelines (Xie et al., 2023) incorporating additional safety exemplars to enable in-context defense (Wei et al., 2023). Wang et al. (2024) introduce a backdoor trigger into safety-aligned LLMs, serving as a covert prefix that elicits safety responses when detected, without affecting model behavior on benign queries. In recent works, Zhang et al. (2024) introduced a dual-stage prompting strategy, Intention Analysis (IA), which encourages LLMs to analyze the intent behind an instruction prior to generating a safe response.

In high-stakes domains such as healthcare and finance, effective decision-making demands not just accurate outcomes but transparent and explainable reasoning. However, current language models often lack the structured deliberation needed for such tasks, instead generating decisions and justifications in a disconnected, post-hoc manner. To address this, we propose DecisionFlow, a novel decision modeling framework that guides models to reason over structured representations of actions, attributes, and constraints. Rather than predicting answers directly from prompts, DecisionFlow builds a semantically grounded decision space and infers a latent utility function to evaluate trade-offs in a transparent, utility-driven manner. This process produces decisions tightly coupled with interpretable rationales reflecting the model's reasoning. Empirical results on two high-stakes benchmarks show that DecisionFlow not only achieves up to 30% accuracy gains over strong prompting baselines but also enhances alignment in outcomes. Our work is a critical step toward integrating symbolic reasoning with LLMs, enabling more accountable, explainable, and reliable LLM decision support systems. Code and data are at https://github.com/xiusic/DecisionFlow.

Large Language Models (LLMs) have demonstrated remarkable capabilities in natural language processing (NLP), including text generation, translation, summarization, and code synthesis, as a consequence of which revolutionizing a wide range of AI applications [10, 56, 45]. Models such as OpenAI's ChatGPT series, Google's Gemini, and Anthropic's Claude have been widely deployed in commercial systems, including search engines, customer support, software development tools, and personal assistants [45, 55, 3]. However, as their capabilities grow, so do their attack surfaces and the potential for misuse [51, 77, 50]. While the scale and specific nature of these vulnerabilities are new, the fundamental challenge of ensuring that powerful AI systems operate safely and align with human intent is a longstanding concern in the AI community. Foundational work, such as the identification of concrete problems in AI safety long before the current LLM era, laid the groundwork for understanding issues like reward hacking and negative side effects that remain highly relevant today [1]. The susceptibility arises because the models are trained on vast, yet imperfectly curated, datasets containing potentially harmful content, and because they interact with users through open-ended prompts that can be manipulated [48, 17, 16]. Researchers and practitioners are increasingly concerned that these systems can be manipulated, misused, or even behave in misaligned and potentially deceptive ways [25, 42, 6]. Consequently, the security and alignment of LLMs have become critical areas of study, requiring an understanding of emergent threats and robust, multi-faceted defenses [17, 70, 43].

Intent detection, a core component of natural language understanding, has considerably evolved as a crucial mechanism in safeguarding large language models (LLMs). While prior work has applied intent detection to enhance LLMs' moderation guardrails, showing a significant success against content-level jailbreaks, the robustness of these intent-aware guardrails under malicious manipulations remains under-explored. In this work, we investigate the vulnerability of intent-aware guardrails and demonstrate that LLMs exhibit implicit intent detection capabilities. We propose a two-stage intent-based prompt-refinement framework, IntentPrompt, that first transforms harmful inquiries into structured outlines and further reframes them into declarative-style narratives by iteratively optimizing prompts via feedback loops to enhance jailbreak success for red-teaming purposes. Extensive experiments across four public benchmarks and various black-box LLMs indicate that our framework consistently outperforms several cutting-edge jailbreak methods and evades even advanced Intent Analysis (IA) and Chain-of-Thought (CoT)-based defenses. Specifically, our "FSTR+SPIN" variant achieves attack success rates ranging from 88.25% to 96.54% against CoT-based defenses on the o1 model, and from 86.75% to 97.12% on the GPT-4o model under IA-based defenses. These findings highlight a critical weakness in LLMs' safety mechanisms and suggest that intent manipulation poses a growing challenge to content moderation guardrails.