Carter Teplica

Hi! I'm an AI safety researcher, currently at MATS doing guardrail design research with LawZero. I'm on leave for the summer from a PhD at Mila and Polytechnique Montréal, studying AI safety under Lina Marsso. I recently did a project with the Alignment Science team at Anthropic. I did a master's at NYU and undergraduate at Columbia. In my free time (among other things) I sing in a choir and do puzzle hunts.

I have a CV.

You can email me by evaluating {first initial}{last name}{at gmail dot com}. You can text me by evaluating +1 {2^3 * 29 * 97 * 233 * 389}. I like talking to people!

Projects

• [Personal project] [2026-01] Seriate: sort your emails by embedding similarity
• [Personal project] [2026-01] train.garden: explore branching transit itineraries
• [Paper (as mentor)] [2025-08] Olivia Long and Carter Teplica. The AI in the Mirror: LLM Self-Recognition in an Iterated Public Goods Game
• [Paper (as joint first author)] [2025-03] Carter Teplica and Hongxin Song.Sociolinguistic Simulacra: Emergent Regional Behavior in Fine-Tuned Language Models
• [Paper (as first author)] [2024-12] Carter Teplica, Yixin Liu, Arman Cohen, and Tim G. J. Rudner. SCIURus: Shared Circuits for Interpretable Uncertainty Representations in LMs
• [Blog post] [2024-04] Singularities and the Edge of Stability
• [Blog post] [2023-08] A Mechanistic Analysis of Counting in Distil-GPT2