Jianfeng Chi's Homepage

Research Scientist
GenAI, Meta

Email: jianfengchi(AT)meta(DOT)com

[Google Scholar] [Twitter] [LinkedIn]

[Meta Homepage] [Github]

  

About Me


I am a research scientist at GenAI Meta, working on LLM safety problems. Previously, I received my Ph.D. in Computer Science from the University of Virginia in 2022. During my Ph.D. study, I also interned at Facebook and AWS. My research interests broadly lie in machine learning and natural language processing, with a specific focus on ML/AI safety, security & privacy, and algorithmic fairness.

Selected Publications / Technical Reports


* indicates equal/core contributions, listed in alphabetical order.
† indicates equal advising.
♦ indicates students or interns I mentored or closely collaborated.

For a full list of my publications, please go to my Google Scholar webpage.


Llama Guard 3 Vision: Safeguarding Human-AI Image Understanding Conversations (Technical Report AI@Meta)
Jianfeng Chi*, Ujjwal Karn*, Hongyuan Zhan*, Eric Smith*, Javier Rando, Yiming Zhang, Kate Plawiak, Zacharie Delpierre Coudert, Kartikeya Upasani†, Mahesh Pasupuleti†
[pdf] [code]

Backtracking Improves Generation Safety (ICLR 2025, Oral)
Yiming Zhang♦, Jianfeng Chi, Hailey Nguyen, Kartikeya Upasani, Daniel Bikel, Jason Weston†, Eric Michael Smith†.
[pdf]

Persistent Pre-Training Poisoning of LLMs (ICLR 2025)
Yiming Zhang*, Javier Rando*, Ivan Evtimov, Jianfeng Chi, Eric Michael Smith, Nicholas Carlini†, Florian Tramèr†, Daphne Ippolito†.
[pdf]

Towards Understanding the Fragility of Multilingual LLMs against Fine-Tuning Attacks (NAACL Findings 2025)
Samuele Poppi♦, Zheng-Xin Yong♦, Yifei He, Bobbie Chern, Han Zhao, Aobo Yang†, Jianfeng Chi†
[pdf]

BadMerging: Backdoor Attacks Against Model Merging (CCS 2024)
Jinghuai Zhang♦, Jianfeng Chi, Zheng Li, Kunlin Cai, Yang Zhang, Yuan Tian.
[pdf][code]

The Llama 3 Herd of Models (Technical Report AI@Meta)
Llama Team, AI @ Meta
Role: core contributor, responsible for system-level safety and help with pre-training + post-training safety
[website][arxiv]

Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations (Technical Report AI@Meta)
Hakan Inan, Kartikeya Upasani, Jianfeng Chi, Rashi Rungta, Krithika Iyer, Yuning Mao, Michael Tontchev, Qing Hu, Brian Fuller, Davide Testuggine, Madian Khabsa.
[pdf] [blog post] [code]

Where have you been? A Study of Privacy Risk for Point-of-Interest Recommendation (KDD 2024)
Kunlin Cai♦, Jinghuai Zhang♦, Zhiqing Hong, William Shand, Guang Wang, Desheng Zhang, Jianfeng Chi, Yuan Tian.
[pdf][code]

FFB: A Fair Fairness Benchmark for In-Processing Group Fairness Methods (ICLR 2024)
Xiaotian Han♦, Jianfeng Chi, Yu Chen, Qifan Wang, Han Zhao, Na Zou, Xia Hu.
[pdf][code]

Publications during my Ph.D. study

PLUE: Language Understanding Evaluation Benchmark for Privacy Policies in English (ACL 2023)
Jianfeng Chi, Wasi Uddin Ahmad, Yuan Tian, Kai-Wei Chang.
[pdf][code]

Retrieval Enhanced Data Augmentation for Question Answering on Privacy Policies (EACL 2023)
Md Rizwan Parvez, Jianfeng Chi, Wasi Uddin Ahmad, Yuan Tian, Kai-Wei Chang.
[pdf]

Conditional Supervised Contrastive Learning for Fair Text Classification (EMNLP Findings 2022)
Jianfeng Chi, William Shand, Yaodong Yu, Kai-Wei Chang, Han Zhao, Yuan Tian.
[pdf][code]

Towards Return Parity in Markov Decision Processes (AISTATS 2022)
Jianfeng Chi, Jian Shen, Xinyi Dai, Weinan Zhang, Yuan Tian, Han Zhao.
[pdf] [code]

Understanding and Mitigating Accuracy Disparity in Regression (ICML 2021)
Jianfeng Chi, Yuan Tian, Geoffrey J. Gordon, Han Zhao.
[pdf] [code]

Intent Classification and Slot Filling for Privacy Policies (ACL 2021)
Wasi Uddin Ahmad*, Jianfeng Chi*, Tu Le, Thomas Norton, Yuan Tian, Kai-Wei Chang.
[pdf] [code] [Video]

Trade-offs and Guarantees of Adversarial Representation Learning for Information Obfuscation (NeurIPS 2020)
Jianfeng Chi*, Han Zhao*, Yuan Tian, Geoffrey J. Gordon.
[pdf] [Poster] [Slides]

PolicyQA: A Reading Comprehension Dataset for Privacy Policies (EMNLP Findings 2020)
Wasi Uddin Ahmad*, Jianfeng Chi*, Yuan Tian, Kai-Wei Chang.
[pdf] [code]

Hybrid Batch Attacks: Finding Black-box Adversarial Examples with Limited Queries (USENIX Security 2020)
Fnu Suya, Jianfeng Chi, David Evans, Yuan Tian.
[pdf] [code]

Professional Services


PC Member/Reviewer:

Conferences:

Journals: