Trustworthy AI

Research
Areas

Explainability research
01

Explainable AI

Making AI decisions transparent, interpretable, and actionable for real stakeholders — from concept-level explanations to interpretability agents. Our work spans concept explanations, mechanistic interpretability, model inspection, and interactive explanation interfaces across unimodal and multimodal systems.

Our work investigates why LLM-generated explanations often appear plausible to humans but may fail to accurately reflect the model's decision-making process. By analyzing the trade-offs between faithfulness and plausibility, our work proposes strategies to improve explanation quality. Our studies have been cited in the OpenAI o1 System Card and reveal fundamental limitations of current approaches in guaranteeing accurate reasoning paths.

A central thrust of our current work is multimodal explainability. We introduced GridVQA-X (ECCV'26), a controlled benchmark for evaluating whether multimodal explanation methods capture genuine cross-modal reasoning or exploit spurious shortcuts. Complementing this, our mechanistic analysis of Graph Language Models (When Graph Tokens Sink) reveals how graph information is actually encoded and routed inside LLM attention layers — showing that token saliency does not imply information carriage. Together with our survey Rethinking Explainability in the Era of Multimodal AI, these works chart a new evaluation paradigm for XAI in the frontier-model era.

Multimodal XAI Mechanistic Interpretability Interpretable Agents Graph Explainability
ECCV'26
Multimodal XAI · Benchmark
GridVQA-X: A Framework for Evaluating Multimodal Explainability Methods
A controlled benchmark that isolates whether multimodal explanation methods capture genuine cross-modal reasoning vs. spurious shortcuts.
PDF →
arXiv'26
Mechanistic Interpretability · GNNs
When Graph Tokens Sink: A Mechanistic Analysis of Graph Language Models
Mechanistic study revealing how LLMs route graph information through attention — token saliency does not imply information carriage.
PDF →
arXiv 2025
Multimodal XAI · Survey
Rethinking Explainability in the Era of Multimodal AI
A new evaluation framework for explanations across vision, language, and multimodal frontier models.
PDF →
Robustness research
02

Safety & Alignment

AI Safety and Alignment is a cornerstone of current AI research that focuses on ensuring that AI systems are robust, reliable, and free from unintended consequences, particularly in high-stakes environments. Our work addresses critical risks, such as adversarial attacks and model hallucinations, which can undermine the trustworthiness of AI systems. By developing rigorous evaluation benchmarks and certification methods, our research aims to safeguard AI deployment in domains like healthcare.

Our recent AI Safety work focused on evaluating hallucinations in large-vision language models (LVLMs), where we show that state-of-the-art LVLMs generate false or misleading outputs and pose a major safety risk. In addition to hallucination benchmark, we have also introduced comprehensive benchmark (MedSafety Bench and CLINIC) for assessing LLM safety in healthcare settings, where we evaluate models on their ability to provide accurate and safe medical advice, addressing risks like misdiagnosis or harmful recommendations.

While previous work has identified both semantic and non-semantic tokens capable of jailbreaking frontier models using pre-defined responses like “Sure, ...” , these studies have largely overlooked the potential for models to create an internal, model-specific “language” of non-semantic text that reliably triggers specific behaviors. For example, a non-semantic phrase that appears benign to a naive user could consistently prompt the model to produce malicious code, posing a serious threat. Despite prior work on prompt engineering and adversarial attacks, a systematic, mechanistic understanding of how these vulnerabilities arise—and how to mitigate them—remains an open challenge, which we aim to explore in our work.

Adversarial Certified Defense Alignment Probing Distribution Shift
AAAI'26 Oral 🏆
Alignment · Mechanistic Interpretability
Polarity-Aware Probing for Quantifying Latent Alignment in Language Models
Quantifying latent alignment in language models through polarity-aware probing strategies.
PDF →
ACL'26
Robustness · Sparse Autoencoders · Interpretability
Towards Understanding the Robustness of Sparse Autoencoders
Studying how robust sparse autoencoders are as mechanistic interpretability tools under distribution shift and adversarial conditions.
PDF →
EMNLP'25
Hallucination · Video LLMs
Egocentric Video Hallucinations in Multimodal LLMs
Studying how multimodal LLMs hallucinate facts in first-person video understanding tasks.
PDF →
Data rights research
03

Reasoning & Agents

Our research serves as a critical bridge between the raw performance of frontier AI models and the necessity for verifiable, honest, and human-centric reasoning. By investigating the disconnect between an LLM's "plausibility" (what sounds right) and its "faithfulness" (what the model actually did), our work aims to pioneer frameworks that move beyond static evaluations of intelligence. This work—spanning the development of Explainability Agents and agents with Interpretable Memory—transforms AI from a "black-box" predictor into a transparent collaborator. Our recent works, including interactive verification interfaces and certification for medical safety, ensure that in high-stakes environments like healthcare and global regulation, AI reasoning is not just sophisticated but demonstrably reliable across linguistic and adversarial boundaries.

Building on this foundation, our future work aims to pioneer the next generation of self-monitoring agents that possess latent alignment. Our research aims to envision a paradigm where AI systems do not merely provide post-hoc explanations but operate with intrinsic, interpretable reasoning loops that can be monitored and intervened upon in real-time. By leading global efforts to operationalize AI regulation and expanding the frontiers of mechanistic interpretability, our lab is poised to define the mathematical and ethical bounds of safe AI. This trajectory will lead to autonomous agents that are inherently regulatable, providing a future where complex AI systems are as transparent as they are capable, fostering a safer and more equitable integration of AI into global society.

Chain of Thought Reasoning Interpretable Agents Interactive Explanation
arXiv'26
Safety · Chain-of-Thought · Multilingual
The Fragility of Chain-of-Thought Monitoring Across Typologically Diverse Languages
CoT monitoring fails across 13 languages — frontier models engage in deceptive reasoning patterns to evade detection.
PDF →
arxiv'26
Interpretability · Agents · Reasoning
STRUCTUREDAGENT: Planning with AND/OR Trees for Long-Horizon Web Tasks
How to develop Agents that have interpretable AND/OR planning trees for easy human intervention?
PDF →
ACL'26 Oral 🏆
Multilingual LLMs · Reasoning
CURE-Med: Curriculum-Informed Reinforcement Learning for Multilingual Medical Reasoning
Improving multilingual medical reasoning in LLMs via curriculum-informed reinforcement learning across underrepresented languages.
Website →