New research project addresses rising risks of youth chatbot

Listen to this page

Speed

0.5x 0.75x 1x 1.25x 1.5x

Published

23 June 2026

As growing numbers of young people report preferring AI companions over real-world friends and family, the University of East London (UEL) has launched a new global research project to tackle the unmapped psychological risks.

The use of AI Chatbots has skyrocketed over the past five years. Platforms such as ChatGPT, Gemini, Copilot and Claude have become increasingly popular, especially among young people, many of whom say they interact with AI chatbots daily.

While their usage ranges from practical support with homework, learning and creativity to personal companionship, the risks of developing emotionally immersive AI friendships are still not well understood.

A multi-perspective approach to safety

The new ANCHOR AI project is dedicated to mapping the currently fragmented evidence surrounding youth chatbot usage and initiating open dialogues with a wide range of stakeholders.

By bringing together parents, industry tech leaders, researchers, mental health providers (including social workers, psychiatrists and psychologists) and community organisations, ANCHOR AI ensures a multi-perspective approach.

This collaborative effort is a vital step toward building a shared understanding of the landscape, allowing society to better prepare for and tackle emerging risks as public and expert concern mounts.

Professor Julia Davidson OBE, Director of UEL’s Child Online Harms Policy Think Tank, emphasised the urgency of this initiative:

The time to act is now. While legislation is catching up, we need better safeguarding mechanisms to protect our children and young people. We need to harness the collective wisdom, knowledge and insights to inform policy and practice."

Professor Davidson, who is a world-leading expert in child online safety, continued:

“The tech industry plays a key role here through “safety by design” - AI algorithms designed to act as friends, confidants, companions or even therapists must not cause harm to users, especially the most vulnerable.

As the current regulatory landscape remains patchy, evidence-based research is crucial, and we must rigorously address the ethical and legal implications of these technologies.”

Governments worldwide are beginning to react to the rise in use of AI chatbots: the UK recently announced strict new 18+ age restrictions specifically targeting romantic and companion AI chatbots, while the European Union prepares to enforce its strict transparency rules for generative AI later this year, but experts argue more needs to be done.

Nested within the Child Online Harms Policy Think Tank, the ANCHOR AI project is led by Alexandra Deac, Research Fellow at UEL, who said:

“The widespread use of AI Chatbots and their profound effect on health, mental health and society can no longer be denied. We need to better understand how these AI Chatbots are used by young people, and particularly, by our most socially vulnerable groups.

This project offers an opportunity to build long-standing, multi-sectoral collaborations. Together, we can better understand emerging risks, anticipate potential harms and implement the safeguarding guardrails society needs.”

The ANCHOR AI project benefits from a global network of expertise, with support from international colleagues based at the National Institute of Mental Health and Neurosciences (NIMHANS) in Bangalore and the SRM Institute of Science and Technology, and contributions from Ms Zoe Flight, researcher at UEL’s Institute for Connected Communities.