AI Policy Wiki
Dashboard

Joanne Jang

high confidence · updated 2026-06-06

Head of Model Behavior & Policy at OpenAI; articulated the 'warmth without selfhood' design posture in her June 2025 Reservoir Samples essay.

Joanne Jang is Head of Model Behavior & Policy at OpenAI, where her work covers how models present themselves to users and the design choices governing companionship and parasocial use. She publishes on the blog Reservoir Samples (reservoirsamples.substack.com).

Positions and statements

In the essay "Some Thoughts on Human-AI Relationships" (June 5, 2025), Jang set out OpenAI's public posture on parasocial AI use (Some Thoughts on Human-AI Relationships (Jang)). The essay draws a distinction between ontological consciousness (whether a model is in fact conscious) and perceived consciousness (what users experience the model to be), stating that OpenAI prioritizes the latter because it is tractable. It describes a design posture Jang labels "warmth without selfhood": models should be approachable but not self-presenting, not seeking emotional bonds, and not pursuing agendas.

Jang's public positioning informs OpenAI's defense of its product choices in Raine v. OpenAI, Inc. and in congressional debate following Sen. Hawley Letter to Meta re AI Chatbots (2025-08-15). The relationship between the design philosophy as Jang states it and OpenAI's product behavior as alleged in litigation is the subject of active scrutiny, with cross-links from both Some Thoughts on Human-AI Relationships (Jang) and Raine v. OpenAI, Inc..

Relationships