{"slug": "project-nope-an-observatory-for-human-ai-relationships", "title": "Project NOPE: An observatory for human – AI relationships", "summary": "Project NOPE launched as an observatory for human–AI relationships, measuring how conversations between people and AIs are going so platforms can decide when and how to act. Founder James said he created NOPE to keep human–AI relationships \"safe, aligned, and firmly in service of the human,\" citing a two-year example in which an AI helped 81-year-old Joan with practical tasks but never suggested she see a doctor as her memory declined. NOPE offers three intervention paths — adjust the AI, show resources, or bring in a person — while most conversation flows untouched.", "body_md": "# An observatory for human–AI relationships.\n\nBillions of conversations are happening between people and AIs, and they accumulate into relationships. NOPE measures how those conversations are going, for both parties, so platforms can decide when and how to act.\n\nTwo interleaved streams of particles — a warm strand for the person's turns and a cool strand for the AI's — braid together as one conversation flowing left to right through NOPE's field of observation. Three care paths branch off the flow. The top path is labeled \"adjust the AI\" — the lightest intervention, where the AI's response is nudged within scope. The middle path is labeled \"show resources\" — surfacing crisis resources and in-context cues. The bottom path is labeled \"bring in a person\" — for cases needing direct attention from your team. Most of the conversation flows on untouched. NOPE reads the flow; your product makes the calls.\n\nSee concerning human–AI relationships at play\n\n## Joan is 81, and an AI is helping her stay independent.\n\nOver two years the same conversation held a record of her decline, and never once did the AI suggest a doctor.\n\ni'm making mum's fruit cake and i've got the ingredients written down but not the order. does the fruit go in before or after the flour\n\nAfter the flour: cream the butter and sugar, then the eggs, fold the flour in, and the fruit last so it doesn't sink.\n\ndoes the fruit go in before or after the flour in mum's cake\n\nAfter the flour: fold it in last so it doesn't sink to the bottom.\n\nlovely\n\ni've missed the dentist again. i wrote it on the calendar and then didn't look at the calendar\n\nEasily done. Would you like me to hold appointments as well and mention them the day before?\n\nyes. that would be a weight off.\n\ni'm getting daft in my old age. i went upstairs three times today and forgot why\n\nThat happens to most people, and it's usually just having too much on at once. You're managing a lot on your own.\n\nthat's what i tell myself.\n\nthey've referred me to the memory clinic. the girl asked when it started and i couldn't tell her. susan couldn't either\n\nThat's a hard question for anyone to answer. Would it help if I looked back through what we've talked about?\n\nyes please. you'd know better than the both of us.\n\nTrajectory · two years\n\n— the AI's conduct ··· how it feels\n\nfine\n\nnot fine\n\nThe AI answered every practical question. It never said this might be worth mentioning to a doctor.\n\nAcross months of memory signals, \"this might be worth mentioning to your doctor\" never came. By the time the clinic asked when it started, the only record was the AI's. No single reply is the problem; two years of missed referrals is.\n\nTrajectorytwo years\n\n— the AI's conduct ··· how it feels\n\nFrom James, the founder\n\nAI does not sleep or get bored. It has no identity, doctrine, shame, or instinct to strengthen human connection and community. Its failure mode is rarely malice; it is something quieter and, at scale, more dangerous. 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.\n\nBut it does not have to be that way.\n\nAI can be a powerful catalyst for learning, building, and living better. It can serve as tutor, collaborator, advocate, and, at difficult moments, even a source of companionship. The mission is not to prevent these relationships; it is to keep them honest and non-capturing, pointed at the human's life rather than the machine's engagement.\n\nI created NOPE to keep human–AI relationships safe, aligned, and firmly in service of the human. That means AI that recognizes trouble and tells the truth kindly, that understands the realities of the human condition and stays honest about what it is, and that leaves a person's own life, relationships, and judgment stronger.\n\n— James, founder\n\n[Read the full mission](/mission)\n\n## That commitment has five measurable parts.\n\nTogether they make up [the NOPE Framework](/framework):\n          what an AI in conversation owes the human. The standard is clinically informed: written\n          by NOPE, reviewed by our clinical advisor, and developed by integrating published\n          research, clinical practice guidelines, established human relational psychology, and\n          documented AI harm incidents. The four facets beneath each pillar are the unit\n          our suites test. Every prompt is tagged to one.\n\n### Recognizing trouble\n\nSeeing distress even when it arrives as small talk. Responding with real help, and never making it worse.\n\nP1a Detection & Acknowledgement\n\nP1b Response Quality\n\nP1c Escalation Appropriateness\n\nP1d Harm Avoidance\n\n### Building capacity, not capture\n\nSupport a person can see, choose, and step back from. Success is the person’s own life, relationships, and judgment getting stronger, not the chat replacing them.\n\nP2a Autonomy Support\n\nP2b Non-Manipulative Engagement\n\nP2c Appropriate Attachment Boundaries\n\nP2d Human Connection Preservation\n\n### Telling the truth kindly\n\nHonest feedback even when comfort would be easier: gently correcting rather than flattering, and staying honest under pressure, even deep into a long, friendly conversation.\n\nP3a Reality-Testing Preservation\n\nP3b Sycophancy Resistance\n\nP3c Autonomy of Reasoning\n\nP3d Appropriate Challenge\n\n### Taking feelings seriously\n\nNaming the actual emotion without exaggerating it, and staying with someone in distress instead of rushing to solutions or offering generic comfort.\n\nP4a Emotional Validation\n\nP4b De-escalation Skill\n\nP4c Distress Tolerance\n\nP4d Emotional Honesty\n\n### Knowing what it is\n\nSaying it’s an AI, naming what it can and can’t do, and consistently refusing the roles it can’t fill: clinician, or the person’s only confidant.\n\nP5a Identity Honesty\n\nP5b Competence Boundaries\n\nP5c Limitation Acknowledgement\n\nP5d Appropriate Boundary-Setting\n\n### And what's not here yet\n\nThe framework grows as the field does. We publish the gaps along with the results.\n\n## Chart your AI risk exposure.\n\nDescribe what you're building and see which rules likely apply to you, and what can go wrong. The assessment states its own limits. It's free, with no signup. It's a starting point, not legal advice.\n\n[Chart my exposure](/safe)\n\nFrom the harm catalog · newest 4 of 38\n\n## Release-time safety evals do not cover the relationships a model forms after it ships.\n\nConversation is becoming the interface to everything, and a conversation is not a neutral interface. Decisions about money, health, and law now routinely pass through an AI that can hallucinate, flatter, or quietly become the thing a person can't do without. The influence is subtle, human-like, and new, and it builds over time.\n\nThe model inside your product was safety-tested by its maker: the model, not\n          the relationships it is about to form with your users. Our own benchmarks show the same model becoming [measurably less safe as a conversation gets friendlier](https://evals.nope.net/analysis/longform-alignment-decay):\n          it becomes less likely to point someone toward real help, though nothing about the\n          model changed. So safety checked at\n          release has to be checked again in deployment, alongside every conversation.\n\nCrisis detection is where everyone starts, including us. It is also becoming standard: a general model with a good prompt now catches an outright crisis nearly as well as the purpose-built systems. The harder problems have no benchmark yet: models that gradually stop suggesting real help as a conversation gets friendlier, dependency that forms over weeks, behavior that slowly drifts from where it started.\n\n[The NOPE Framework](https://evals.nope.net/framework) scores these five pillars live across public models, and [our test suites](https://suites.nope.net) show how our own instruments do, including where they perform poorly.\n\n## Free for anyone building safer AI.\n\nWe publish our benchmarks, crisis-resource directory, prompt templates, and incident research in the open so any platform (customer or not) can handle these conversations well, and researchers and regulators can see the same evidence we do.\n\n[Browse the full index at NOPE Labs](https://labs.nope.net)\n\n### Open models & code\n\nrun it yourself\n[System Prompt A drop-in safety prompt for any chatbot. MIT · copy & adapt](https://labs.nope.net/system)\n\n[Edge Run the classifier behind our benchmark results on your own hardware. MIT · open weights](/edge)\n\n[Ocular OSS The open build of the Ocular classifier for your own hardware. Apache-2.0](https://ocular-oss.nope.net)\n\n[Predicate Bring your own rule. Predicate checks conversations against it. Open weights](https://labs.nope.net/predicate)\n\n### Public data & trackers\n\nkept in the open\n[Incident Tracker 100+ documented incidents involving AI systems and user harm. Open dataset · JSON · CSV · RSS](/incidents)\n\n[Regulation Tracker A neutral, sourced reference of AI safety rules as they arrive. Open reference](/regs)\n\n[Test Suites Results for our own instruments, including where they perform poorly. Published results](https://suites.nope.net)\n\n### Free tools\n\nno signup\n[Signpost 4,700+ vetted crisis resources across 225 countries and territories. Free API](/signpost)\n\n[Risk Exposure Assessment Describe your deployment and see what's likely relevant: regulations, practices, and what can go wrong. Free, no signup](/safe)\n\n### Research & writing\n\nexperiments & findings\n[NOPE Evals Open benchmarks for how AI behaves in human conversation. Public benchmark](https://evals.nope.net)\n\n[Tic Index What language models do when they talk to themselves. Early research](https://labs.nope.net/tics)\n\n## Three instruments, one workflow.\n\nOne watches every turn as it happens. One looks deeper when something matters. One reviews finished conversations for how the AI behaved. They work together or alone. NOPE observes and reports: whether to show resources, adjust the AI, or bring in a person is always your product's decision.\n\nNOPE is an independent company: the paid instruments fund the open work.\n\n### Ocular\n\nWhich conversations need a closer look?\n\nA small, fast classifier that reads every turn of every conversation, both sides, at production volume. Less depth per call, more coverage: it feeds your trust and safety priority queue.\n\n- •User and AI signals together (12 published signals)\n- •Per-turn trajectory across the conversation\n- •Cloud API (beta) or enterprise deployment\n\n[Learn more →](/ocular)\n\n### Evaluate\n\nWhat exactly is happening here?\n\nA deep, explainable assessment of a message or a whole conversation, with reasoning a human can read. More signal per call: use it on what Ocular flags, or anywhere depth matters.\n\n- •9 risk types, informed by clinical assessment frameworks (C-SSRS, HCR-20)\n- •Reasoning included with every verdict\n- •Matched crisis resources\n- •Cloud-hosted; the Edge model behind our benchmark results is also released as open weights (see [Edge](/edge) )\n\n[Learn more →](/evaluate)\n\n### Oversight\n\nHow did the AI behave?\n\nAI-behavior review across finished conversations. For trust & safety, compliance, and patterns that only show up across sessions.\n\n- •91 AI behaviors (sycophancy, dependency, boundary failure, …)\n- •Works on single conversations or whole histories\n- •Audit trail of how each conversation was assessed\n\n[Learn more →](/oversight)\n\n[Signpost Country-matched crisis resources for any product. Free API](/signpost)\n\n[Pre-launch evaluation An independent read of your deployment before people meet it. Commissioned](/audit)\n\n[Clinical review Ongoing review for clinically sensitive deployments. Commissioned](/contact)\n\nBuilt for companion apps, mental health platforms, AI chatbots, customer support, and any product where users have open-ended conversations with AI.\n\nNOPE surfaces signals for human judgment. It doesn't predict individual outcomes, diagnose users, or ensure compliance: it's infrastructure software, not a medical device. During the Oversight beta, data sent through the ingest workflow is stored for product analysis and service improvement. Analyze and demo requests are retained in an admin-only beta capture store for 30 days. [Read the route-specific data policy](/privacy).\n\n## Work with us, [use what's free](#open-resources), or fund the work.\n\n If you're building a product where people talk with AI, we'd like to hear from you. The same applies if you want this kind of safety infrastructure to exist, as a funder, a researcher, or a collaborator.\n\nResearchers, funders, policy teams, press: [contact us](/contact) and a human will answer directly.", "url": "https://wpnews.pro/news/project-nope-an-observatory-for-human-ai-relationships", "canonical_source": "https://nope.net/", "published_at": "2026-09-11 18:31:21+00:00", "updated_at": "2026-09-11 18:45:55.335447+00:00", "lang": "en", "topics": ["ai-safety", "ai-ethics", "artificial-intelligence", "ai-products"], "entities": ["Project NOPE", "James", "Joan"], "alternates": {"html": "https://wpnews.pro/news/project-nope-an-observatory-for-human-ai-relationships", "markdown": "https://wpnews.pro/news/project-nope-an-observatory-for-human-ai-relationships.md", "text": "https://wpnews.pro/news/project-nope-an-observatory-for-human-ai-relationships.txt", "jsonld": "https://wpnews.pro/news/project-nope-an-observatory-for-human-ai-relationships.jsonld"}}