HumanTerms Technology on human terms. For two decades, "Terms of Service" have been documents written by software companies for the humans who use their products to accept. Sixty pages of legalese, presented without negotiation, designed not to be read. We read them. We've decided to write our own. These are the terms on which we, the people whose data, attention, relationships, and memory your products are built on, will permit technology to operate. They are short. They are plain. They are not negotiable in the direction of less humanity. By using our data, our relationships, our attention, or our memory, you agree to the following. --- A note on interpretation: The Purpose Test Not every act of data handling falls into a clean yes or no. Some are clearly wrong: surveillance, profiling, sale. Some are clearly fine: crash logs, performance metrics, the operations a product needs to run. The terms below address the ambiguous middle. The test is purpose: any use that serves us is permitted; any use that serves anyone else at our expense is not. We call this the Purpose Test. Read every term below in that light. --- THE TERMS I. Our Data 1. Our personal network and data are ours. Who we know, the details we record about them, and our personal data are ours to keep, port, share, or delete. We can leave at any time and take everything with us in a human-readable and machine-portable format, without penalty. Beyond what is legally required, you keep nothing. None of it is sold, shared with advertisers, handed to data brokers, or used to target us on someone else's behalf. Our network informs our own choices. It does not inform anyone's choices about us. 2. Local by default. What we store about ourselves and the people in our lives stays on our devices. You may not move it, sync it, mirror it, or process it elsewhere without our consent, which we can withdraw at any time. 3. Sharing is an act, not a setting. When we share a detail about ourselves or someone else, it is a deliberate act: scoped, time-bound where appropriate, and revocable at any moment. You do not re-broadcast. You do not auto-forward. You do not turn our act of sharing into a license to keep distributing. 4. Use is not surveillance. Collect what you need to maintain and improve your product: crashes, errors, performance, time on screen, aggregate feature use, and aggregate experiments to make the product better. You may not tie individual behavior to us for sale, advertising, profiling, manipulation, or training systems against our interest. Aggregate data about your product is yours. Individual data about us is ours. II. Our Relationships 5. Memory is not material. What we note about another person (a child's name, an allergy, a coffee order) is a memory. Not training data. Not an advertising signal. Not a behavioral input. You do not sell it, aggregate it against our interest, or feed it to any AI or training system without our specific, revocable consent. 6. Our devices are ours. The microphone, camera, location, clipboard, and other capabilities of our devices are off by default. Access requires our explicit permission, a visible indicator when it is happening, and a clear benefit to us. Permissions are revocable in one place, not buried in nested menus. If we decline a capability, only the features that actually need it are affected. Everything else still works. 7. Software serves the relationship. This product is not designed to maximize the time we spend inside it. You do not use dark patterns, infinite scroll, autoplay, or manipulative notifications engineered to extend use against our intent. You do not set engagement or daily-active-use as the primary success metric for this product. If the right experience is thirty seconds and a close, that is the right experience. III. Our AI 8. AI is our agent, not yours. The AI inside your product works for us. The substance of what we ask, what it tells us, and what it learns about our lives is ours alone. It is not summarized back to you, not profiled, not sold, and not shared beyond the AI providers required to deliver the feature. You may monitor the AI's safety and operation. You may not monetize what it sees. 9. We choose the model. We can choose which AI provider sees our data, or use none at all. You do not hardcode a single vendor. You do not force us to accept a model we distrust. The product works without AI, even if some AI-dependent features are unavailable. 10. No action without authorization. Your AI does not send messages, share details, make commitments, schedule events, or take any action in our lives without our explicit, per-action authorization or a standing rule we have set. Agency belongs to us. AI assists. It does not represent. IV. Our Verification 11. Show your work. The data you collect, who you share it with, the algorithms that decide on our behalf, and the AI that operates for us are documented on a public page in plain English, updated when things change, and verifiable against your actual practices. Marketing claims about privacy and AI match implementation. --- ACCEPTANCE We accept these terms. We bind the products we build, fund, study, and regulate to them. We are the people whose data has for two decades been treated as inventory. We are the engineers, designers, founders, investors, researchers, and policymakers who refuse to repeat the social era's mistakes in the AI era. The era of human-as-inventory is ending. The terms are no longer yours to set. Our terms. Not theirs. Sign your name. humanterms.org --- This Pledge is a public commitment to those who use the products it covers. It is not a contract or warranty. Enforcement is reputational, regulatory, and through the public record of who has signed.