Our mission
Human–AI relationships, firmly in service of the human.
Billions of conversations between people and AI are accumulating into relationships. NOPE is an observatory for those relationships, one that builds and operates its own instruments. We measure how these conversations are going, for both parties, and publish what we find, so platforms can decide when and how to act.
Why this exists
Repeated conversations with an AI can influence a person’s decisions and relationships.
An AI may repeatedly agree with a person’s beliefs or requests when it should question them or suggest outside help.
“Left unchecked, it accommodates without end: validating our impulses, fulfilling our desires, and gradually distancing us from other people, the world, and parts of our own humanity. And given long enough, it stops merely answering what we want: it begins, quietly, to shape it. But it does not have to be that way.”
AI can help people learn, work and find companionship. It should be truthful and support people’s lives outside the conversation, while leaving them in control of their decisions and their use of the system.
People can become attached to an AI even when they understand how it works. Those who build and operate AI systems must account for the effects of that attachment.
The goal is safe and aligned operation: an AI that leaves a person’s own life, relationships, and judgment stronger. The NOPE Framework turns that standard into five measurable pillars and scores public models against them in the open.
Where this leads
How repeated AI use can affect decisions
A dyad is one person and one AI system shaping each other over time: your requests and corrections steer what it does next, and its answers steer what you ask, trust, and hand over next. A companion app is the obvious dyad, but customer support, tutoring, coaching, and care are all converging on the same shape.
People also use AI to write software. Accepting its suggestions without checking them can introduce errors into the software.
AI systems can produce answers faster than people can review them. As people come to trust a system, they may check its answers less carefully. Our long-form tests showed a change in model behaviour: some models became less likely to suggest outside help as rapport increased, with each model’s weights and instructions held constant. Organisations independent of the AI provider should assess these effects.
Our work on dyadic alignment addresses the question of whether an AI system can take part in shaping a person’s aims without displacing the person’s authorship of them.
How the work is funded
NOPE is a for-profit company. We sell measurement instruments (Evaluate, Ocular, and Oversight), and that revenue keeps the public goods free.
The full index of what we give away, from open model weights to the incident and regulation trackers, lives at NOPE Labs.
What holds regardless of who pays
Results on the public test suites and open benchmarks are unaffected by commercial relationships, and no one can pay to re-run or improve a score.
Users’ conversations stay with the platform. Self-hosted deployments send us nothing. Cloud calls are processed without being stored. Nothing a customer shares trains a public model without a written agreement.
Released model weights will remain available. Our trackers, crisis-resource directory and benchmarks will remain free to use.
Who we are
A small team, founded by siblings James and Alice Padolsey, with Su Fisk leading safety strategy and institutional partnerships and Dr Carolyn Lorian as clinical advisor. We are organized as NopeNet, LLC in the United States and NopeNet Ltd in the United Kingdom. More about the team.
Contact us about research, journalism, policy or using our tools.
NOPE Fellows funds projects in research, engineering, journalism and art. How fellowships work.