Publications

Conference papers

AgentAbstain: Do LLM Agents Know When Not to Act?

Xun Liu†, Yi Evie Zhang†, Vira Kasprova*, Parisa Rabbani*, Pardis Sadat Zahraei*, Tianyu Zhang*, Ali Ebrahimpour-Boroojeny, Varun Chandrasekaran († project lead; * equal contribution)

NeurIPS 2026 · Oral Advances in Neural Information Processing Systems, 2026, Oral (0.40%)

A paired-task benchmark of 263 should-act / should-abstain tasks in 42 sandboxed tool environments; the best of 17 frontier LLMs gets only 59.5% paired accuracy.

Paper Code Website

@inproceedings{liu2026agentabstain,
  title     = {{AgentAbstain}: Do {LLM} Agents Know When Not to Act?},
  author    = {Liu, Xun and Zhang, Yi Evie and Kasprova, Vira and Rabbani, Parisa and Zahraei, Pardis Sadat and Zhang, Tianyu and Ebrahimpour-Boroojeny, Ali and Chandrasekaran, Varun},
  booktitle = {Advances in Neural Information Processing Systems},
  year      = {2026},
  eprint    = {2607.10059},
  archivePrefix = {arXiv}
}

Too Polite to Disagree: Understanding Sycophancy Propagation in Multi-Agent Systems

Vira Kasprova*, Amruta Parulekar*, Abdulrahman AlRabah*, Krishna Agaram*, Ritwik Garg, Sagar Jha, Nimet Beyza Bozdag, Dilek Hakkani-Tür (* equal contribution)

SIGDIAL 2026 Proceedings of the 27th Annual Meeting of the Special Interest Group on Discourse and Dialogue (SIGDIAL 2026), 2026

Telling agents how sycophantic their peers are curbs error cascades in multi-agent discussion and raises final accuracy by 10.5 points.

Paper Code

@inproceedings{kasprova2026polite,
  title     = {Too Polite to Disagree: Understanding Sycophancy Propagation in Multi-Agent Systems},
  author    = {Kasprova, Vira and Parulekar, Amruta and AlRabah, Abdulrahman and Agaram, Krishna and Garg, Ritwik and Jha, Sagar and Bozdag, Nimet Beyza and Hakkani-T{\"u}r, Dilek},
  booktitle = {Proceedings of the 27th Annual Meeting of the Special Interest Group on Discourse and Dialogue},
  pages     = {795--814},
  year      = {2026},
  publisher = {Association for Computational Linguistics},
  url       = {https://aclanthology.org/2026.sigdial-1.56}
}

Preprints

Sycophancy in Language Models: A Survey Across Behaviors and Topologies

Vira Kasprova, Jingrui He, Dilek Hakkani-Tür, Volodymyr Kindratenko

Under review · ACM CSUR Preprint, under review at ACM Computing Surveys, 2026

Sycophancy is a family of compliance behaviors whose form depends on the interaction's structure; a survey of 126 papers across single-turn, multi-turn, multi-agent, and tool-use settings.

Paper PDF Website

@misc{kasprova2026sycophancy,
  title     = {Sycophancy in Language Models: A Survey Across Behaviors and Topologies},
  author    = {Kasprova, Vira and He, Jingrui and Hakkani-T{\"u}r, Dilek and Kindratenko, Volodymyr},
  year      = {2026},
  publisher = {Zenodo},
  doi       = {10.5281/zenodo.21347166},
  note      = {Under review at ACM Computing Surveys}
}