Faculty Profile: Prof. Daphne Ippolito — CMU LTI

Position: Assistant Professor (joined ~2023) Institution: Carnegie Mellon University, Language Technologies Institute Website: https://www.daphnei.com/ Report date: 2026-06-12

Research Focus

LLM safety, data poisoning / training data contamination, text detection (AI-generated text identification), memorization in LLMs, privacy in generative models.

Academic Profile

PhD: University of Pennsylvania
Previously at Google Brain / Google Research
Research on detecting AI-generated text and understanding what LLMs memorize from training data

Key Publications

Paper	Venue	Focus
Detecting AI-generated text	Various	Distinguishing human vs. model output
Data poisoning in LLMs	2023–2024	Training data contamination and defenses
Memorization in language models	Various	Privacy and data leakage

Fit with Weijia Zhang

Dimension	Assessment
LLM safety / data integrity	✅ Core focus
AI-generated text detection	✅ Strong
Privacy in LLMs	✅ Active area
NLP agents	❌ Not primary focus
GUI / VLM agents	❌ Not her area
RL post-training	❌ Not primary

Verdict

P3 套磁（若做 LLM safety/data integrity 方向）。 Only relevant if Weijia pivots toward LLM safety, data contamination, or evaluation integrity. Not aligned with agentic AI primary direction.