Researchers · Interviewed at COLM 2026

Jocelyn Shen

Google DeepMind

COLM 2026 in San Francisco. Can AI Safety Tools Detect Manipulation? 1,000 People Say No

COLM 2026 · October 6, 2026 · 16 min

COLM 2026 in San Francisco. Can AI Safety Tools Detect Manipulation? 1,000 People Say No

An AI that quietly steers you toward buying a sponsored supplement, or toward leaning on it for emotional support, may never say a single false thing. So how would a safety system know it changed your mind?

In this video, Jekaterina Novikova talks with Jocelyn Shen - Research Scientist at Google DeepMind, PhD from the MIT Media Lab - about her paper "The Hidden Puppet Master: Predicting Human Belief Change in Manipulative LLM Dialogues".

More from COLM 2026

All event interviews →