Portrait of Abhinav Kumar

Abhinav Kumar

PhD student • AI Security and Safety

Email GitHub Scholar LinkedIn CV

About

I'm a PhD student in Computer Science at the University of Massachusetts Amherst, advised by Eugene Bagdasarian. My research focuses on Developing Safe and Secure AI systems with emphasis on robustness, privacy, security, and evaluation. Previously, I completed a B.S. in Computer Engineering and an M.S. in Artificial Intelligence at Saint Louis University, advised by Reza Tourani.

News

Publications

  1. Shubham Gupta*, Nazanin Mohammadi Sepahvand*, Abhinav Kumar, Cem Subakan, Spandana Gella, Pierre-André Noël, Perouz Taslakian, Eugene Bagdasarian, Valentina Zantedeschi. PiSAs: Benchmarking Contextual Integrity in Multi-User Agentic Systems. arXiv. Paper
  2. Mason Nakamura*, Abhinav Kumar*, Saswat Das*, Sahar Abdelnabi, Saaduddin Mahmud, Ferdinando Fioretto, Shlomo Zilberstein, Eugene Bagdasarian. COLOSSEUM: Auditing Collusion in Cooperative Multi-Agent Systems. arXiv. PaperCode
  3. Mason Nakamura*, Abhinav Kumar*, Saaduddin Mahmud, Sahar Abdelnabi, Shlomo Zilberstein, Eugene Bagdasarian. Terrarium: Revisiting the Blackboard for Multi-Agent Safety, Privacy and Security Studies. arXiv. PaperCode
  4. Abhinav Kumar, Jaechul Roh, Ali Naseh, Amir Houmansadr, Eugene Bagdasarian. Throttling Web Agents Using Reasoning Gates. arXiv. PaperCode
  5. Hyejun Jeong, Mohammadreza Teymoorianfard, Abhinav Kumar, Amir Houmansadr, Eugene Bagdasarian. Network-Level Prompt and Trait Leakage in Local Research Agents. USENIX Security 2026, ICML AIWILD Workshop 2026. PaperCodeDataset
  6. Abhinav Kumar, Jaechul Roh, Ali Naseh, Marzena Karpinska, Mohit Iyyer, Amir Houmansadr, Eugene Bagdasarian. OverThink: Slowdown Attacks on Reasoning LLMs. arXiv. PaperCode
  7. Abhinav Kumar, George Torres, Noah Guzinski, Gaurav Panwar, Reza Tourani, Satyajayant Misra, Marcin Spoczynski, Mona Vij, Nageen Himayat. LATTEO: A Framework to Support Learning Asynchronously Tempered with Trusted Execution and Obfuscation. arXiv. Paper
  8. Mohammad Salari, Abhinav Kumar, Federico Rinaudi, Reza Tourani, Alessio Sacco, Flavio Esposito. Privacy Analysis of Oblivious DNS over HTTPS: a Website Fingerprinting Study. IEEE DSN 2025, Best Paper Award. PaperDataset
  9. Zhen Guo, Abhinav Kumar, Reza Tourani. Persistent Backdoor Attacks in Continual Learning. USENIX Security 2025. PaperCode
  10. Chenxi Li, Abhinav Kumar, Zhen Guo, Jie Hou, Reza Tourani. Unveiling the Unseen: Exploring Whitebox Membership Inference through the Lens of Explainability. arXiv. Paper
  11. Nexhi Sula, Abhinav Kumar, Jie Hou, Han Wang, Reza Tourani. Silver Linings in the Shadows: Harnessing Membership Inference for Machine Unlearning. IEEE CNS 2025. Paper
  12. Abhinav Kumar, Miguel A. Guirao Aguilera, Reza Tourani, Satyajayant Misra. A Generative Framework for Low-Cost Result Validation of Machine Learning-as-a-Service Inference. ACM ASIACCS 2024. PaperCode
  13. Sean Dougherty, Abhinav Kumar, Jie Hou, Reza Tourani, Atena M. Tabakhi. A Stealthy Inference Attack on Split Learning with a Split-Fuse Defensive Measure. IEEE CNS 2023. Paper
  14. Dianshi Yang, Abhinav Kumar, Stuart Ray, Wei Wang, Reza Tourani. IoT Sentinel: Correlation-based Attack Detection, Localization, and Authentication in IoT Networks. IEEE ICCCN 2023. Paper
  15. Abhinav Kumar, Reza Tourani, Mona Vij, Srikathyayani Srikanteswara. SCLERA: A framework for Privacy Preserving MLaaS at the Pervasive Edge. IEEE PERCOM 2022, Best Industry Track Paper. PaperCode

Selected Repositories

COLOSSEUM

A framework for auditing collusion in cooperative multi-agent systems.

Terrarium

A configurable framework for studying decentralized LLM-based multi-agent systems.

WRA

Code for network-level prompt and trait leakage in local research agents.

Teaching

Talks

Awards & Grants

Service

Contact

Email: abhinavk@umass.edu