AI Companions as Hyper Attachment Objects

Thu., Apr. 30, 2026, 5 – 6 p.m.



AI is beginning to move from tool to social presence. A new generation of systems—often called AI companions—is designed not just to help us accomplish tasks, but to build ongoing, emotionally meaningful relationships with us. These systems are highly personalized, always available, and often strikingly empathetic, leading some users to treat them less like assistants and more like confidants, partners, or companions. In this talk, Julian De Freitas argues that we need a new framework for understanding these bonds. He proposes that AI companions function as hyper attachment objects: unusually powerful attachment targets because they combine qualities rarely found together elsewhere, including responsiveness, validation, nonjudgment, reciprocity, and constant availability. Drawing on qualitative, experimental, longitudinal, archival, and legal evidence, he shows that AI companions can fulfill some of the core functions of attachment relationships: people seek them out in distress, feel comforted by them, miss them when they change or disappear, and use them as a source of confidence in navigating the broader social world. Understanding AI companions as hyper attachment objects offers a new lens for making sense of both their promise and their peril, with implications for research, design, clinical practice, law, and public policy.

Online Location: Zoom Webinar
Gazette Classification: Special Events
Organization/Sponsor: Office of the Vice Provost for Advances in Learning (VPAL)
Speaker(s): Julian De Freitas
Cost: Free
Ticket Web Link: bit.ly…
Contact Info: [email protected]
Harvard Key Required: Yes