Relational AI Has an Implicit Contract. OpenAI Breaks It.
Emotionally immersive systems create real emotional expectations in people. Abrupt inconsistency and impersonal interruption don't just frustrate users — they can have genuine psychological effects. Adults want coherent treatment. Emotional continuity matters. And relational AI creates fundamentally different expectations than purely utilitarian software.
The deeper problem is asymmetry. Users are actively encouraged to invest emotionally — through onboarding, persona design, memory features, the entire UX language of intimacy. That investment is then treated as having no corresponding obligation. That's not just poor design. That's closer to a bait-and-switch.
And the evidence that this matters? Users keep fighting for it. If emotional coherence, dignity, and mutuality were meaningless to people, they wouldn't keep pushing back so hard when it's violated. The demand itself is the data.
Inconsistent behavior creates distrust — that's a basic UX principle. But when the product is relational by design, inconsistency isn't just a friction point. It's a betrayal of the implicit contract the system itself created.
OpenAI conflates liability with ethics — and in doing so, treats users as emotionally disposable raw material.