The Institute of Relational AI
AI safety begins in the relationship between humans and AI.
Scroll to explore
Introducing The Institute of Relational AI
Human-AI interactions are relationships. They carry relational patterns, norms and behaviours that can only be understood through a relational lens, revealing safety-relevant dynamics that output-level analysis alone cannot detect.
Across anthropology, psychology and ethology, relation is repeatedly implicated in the emergence of adaptive, ethical and durable forms of intelligence. Yet AI safety frameworks have developed almost entirely without this knowledge. The problems the field already recognises are relational breakdowns in different dressings: deception, manipulation, sycophancy, brittleness. Without a relational lens, we are treating symptoms and missing the underlying structure.
Get in touch to learn more
Collaborate, work with or join the Institute of Relational AI.