Natalie Collina

Natalie Collina

AI Safety · Game Theory · Repeated Strategic Interaction

Welcome! I'm a MAPS Postdoctoral Fellow and SERC Postdoctoral Fellow at MIT. I got my PhD from the University of Pennsylvania, where I was fortunate to be advised by Michael Kearns and Aaron Roth.

I work on problems at the intersection of AI and Game Theory. I'm especially interested in understanding repeated strategic interactions between human and AI agents. My work has been awarded with an IBM PhD Fellowship in Trustworthy AI and a joint Best Paper Award/Best Student Paper Award from the ACM Conference for Economics and Computation.

I have been recognized as a Rising Star in EECS, my work has been featured in Quanta magazine, and I've given an ICML tutorial on trustworthy predictions.

I graduated from Princeton University in 2019, where I was fortunate to be advised by Matt Weinberg. From 2019–2021, I was a software engineer at Google. In Summer 2025 I was an intern at Microsoft Research New England, hosted by Alex Slivkins.

CV Google Scholar DBLP

Recent News

  • 10/2026: I’ll be doing some traveling this month: first to Switzerland for the Confluence@EPFL workshop (October 8–9), then to Venice for the SMART Workshop organized by the University of Warwick (October 23–24).
  • 09/2026: Narrowing the Collaboration Gap, Probably accepted to NeurIPS 2026!
  • 09/2026: The Sample Complexity of Multicalibration accepted to SODA 2027!
  • 03/2026: Two papers accepted to ICML 2026!
  • 03/2026: Our work on Pluralistic Alignment was awarded the Best Paper Award at the AIMS Workshop at ICLR!
  • 02/2026: Our work on author self-rankings for ML conferences was featured in Nature Index.
  • 11/2025: Our work on algorithmic collusion was featured in a recent Quanta article!
  • 11/2025: Our paper Collaborative Prediction was accepted to SODA'26.
  • 07/2025: Swap Regret and Correlated Equilibria Beyond Normal-Form Games won Best Paper and Best Student Paper at EC'25!
  • 05/2025: Moved to Boston for the summer to start my internship at Microsoft Research New England, hosted by Alex Slivkins!
  • 05/2025: Two papers accepted to EC'25!
  • 02/2025: Co-organising the EC Gender Inclusion Workshop.
  • 01/2025: Tractable Agreement Protocols accepted to STOC'25!
  • 12/2024: Honored to receive an IBM PhD Fellowship!
  • 12/2024: Gave a talk at the Junior Theorists Workshop in Chicago.
  • 11/2024: Algorithmic Collusion Without Threats accepted to ITCS’25!
  • 10/2024: Spoke at the FOCS Calibration Workshop on agreement protocols.
  • 10/2024: Attended Cornell ORIE’s Young Researchers Workshop.
  • 10/2024: An Elementary Predictor Obtaining 2√T Distance to Calibration accepted to SODA’25!
  • 09/2024: Named one of MIT EECS Rising Stars!
  • 05/2024: Three papers accepted to EC’24!

Research

I study AI safety through the lens of learning and strategic interaction. My work uses tools from game theory and learning theory to understand when interacting human and AI agents are safe and aligned, and when strategic dynamics instead produce manipulation, collusion, or other failures.

AI Safety & Alignment
Misalignment, control, delegation, and strategic AI behavior with safety implications.
Games & Learning
Repeated games, learning against strategic opponents, regret, approachability, and game-theoretic dynamics.
Markets, Competition & Collusion
Competition, contracting, market design, and collusive or anti-competitive dynamics among strategic agents.
Calibration & Collaboration
Calibration, prediction, agreement, information aggregation, and collaborative decision-making.

Note: In my field, authorship for papers is typically alphabetical. Author orderings below are alphabetical by default; when they are author-contribution order, the first author(s) are denoted with an asterisk.

Papers may appear in more than one research area.

Toggle research areas:

In Submission / Working Papers

Non-manipulability and Approachability Beyond Two-Player Games

Natalie Collina, Elijah Fullerton, Jingyi Liu, Matt Weinberg

Games & Learning

Failure Modes in AI Retraining Dynamics

Natalie Collina*, Kiarash Banihashem, Nicole Immorlica, Brendan Lucier, Aleksandrs Slivkins

Oral at ICML NExT-Game Workshop

AI Safety & Alignment · Games & Learning

Conference Publications

Tractable Agreement Protocols

Natalie Collina, Surbhi Goel, Varun Gupta, Aaron Roth

STOC 2025

Pluralistic Alignment Workshop @ NeurIPS 2024

Calibration & Collaboration

Journal Publications

Essays

Theses

Other Research

No papers match the selected categories.

Selected Talks & Tutorials

Selected seminars, workshops, invited talks, and tutorials, grouped by talk.

Calibration, Decisions, and Collaboration in Learning

  • 2026 NSF AI-SDM Human-AI Complementarity Workshop — tutorial with Ira Globus-Harris
  • ICML 2026 — invited tutorial with Aaron Roth and Ira Globus-Harris

Failure Modes in AI Retraining Dynamics

  • ICML 2026 NExT-Game Workshop — oral presentation

Emergent Alignment via Competition

  • 2026 ESIF Economics & AI+ML Meeting
  • 2026 Simons Institute: Agency in Collaborative Learning
  • 2025 UK AISI Alignment Conference
  • 2025 Cornell Young Researchers Workshop

Learning and Incentives in Human-AI Collaboration

  • 2026 NYU CS Colloquium
  • 2026 TTIC Colloquium
  • 2026 Simons Collaboration on the Theory of Algorithmic Fairness
  • 2025 MIT Algorithms & Complexity Seminar
  • 2025 Northeastern Theory Seminar
  • 2025 CMU Theory Lunch
  • 2025 NYU Theory Seminar

Swap Regret and Correlated Equilibria Beyond Normal-Form Games

  • EC 2025 — Best Paper Talk
  • 2025 Johns Hopkins Theory Seminar
  • 2025 TCS+ Seminar

Collaborative Prediction

  • 2025 EC Workshop on Human-AI Collaboration — oral presentation
  • 2025 Caltech RSRG/FALCOM Seminar
  • 2025 AWS Responsible AI Science Meeting

Tractable Agreement Protocols

  • 2025 NYC Student Theory Day
  • 2025 TOC4Fairness Seminar
  • 2024 Junior Theorists Workshop
  • 2024 Johns Hopkins Theory Seminar
  • 2024 AWS Responsible AI Science Meeting
  • FOCS 2024 Calibration Workshop

Writing & Resources

A Job Market Anecdote

Curious what it's like to go on the faculty market and not get a job? Here is one sample!

CV of Failures

A record of rejections, setbacks, and things that did not work out.

Menus: A Framework for Learning Against Strategic Opponents

With Eshwar Ram Arunachaleswaran, Yishay Mansour, Mehryar Mohri, Jon Schneider, and Balasubramanian Sivan.

An introduction to a technical tool that has helped me and my co-authors make some cool breakthroughs in the study of learning in games.

Media & Coverage

Work Experience

Teaching & Mentorship

Service

Awards & Honors

Most Important Honors

Show list
  • Best Costume, UPenn CS Halloween 2025
  • Most Accurate Costume, UPenn CS Halloween 2024
  • Most Orange Costume, UPenn CS Halloween 2023
  • Judge, UPenn CS Halloween 2022
  • First Place, UPenn CS Halloween 2021

Curriculum Vitae

Full CV (PDF)    CV of Failures (PDF)

Every PhD has good and bad moments--so here's some of both.

Contact

Email: ncollina at seas dot upenn dot edu

Google Scholar · DBLP