Your Model, Your Terms

A few years ago, researchers built a fake social network and invited 543 people to join. Buried in the fine print of the terms and conditions was a clause handing their data to the NSA and to their employer, and another requiring them to give up their first-born child as payment for the service. Ninety-eight percent missed both. Three-quarters never opened the document at all. The ones who did read it spent about a minute on something that if read in full, would have taken over an hour.

Every day, your life is shaped by terms and conditions you have never read and never will. They are a standard part of using every app, agent, and AI you depend on. While most terms and conditions don’t require handing over your first born in exchange for using a product, this study proved a point. We have no choice but to accept the terms we’re offered, because we could never be expected to read them, and even if we did, we’d have no leverage to change them.

This sort of take-it-or-leave-it expectation is one thing when you're ordering a pair of shoes or downloading a food delivery app. The terms cover one purchase or one app, and the stakes are lower. A model of your life is different. It knows how you think, it influences what you see and do, and it sits between you and everything else you use.

In just the past few weeks, River AI raised over a billion dollars for personal AI, Instinct raised $250 million for a personal assistant designed to understand and act for you, and Mark Zuckerberg published a manifesto describing his vision for personal superintelligence. The category is emerging because powerful open source models can now run locally on your laptops and phones. These models will power agents, personal assistants, and eventually robots, using our data and the signals our devices emit to understand us. Tech companies promise that these models will be private, that they won’t be able to see in them, that they won’t be survielling you. I’ll take them at their word, but frankly, privacy isn’t enough. And open source models don’t inherently solve the problem either. They still come with terms and conditions for products built on top of them and are not easy to install or manage.

You see, the terms and conditions that these companies write, and the leverage they have to get us to accept them nearly blindly, means the most personal and influential product in your life, your personal model, won’t be fully controlled by you.

But why couldn’t you have your own terms, set by you, based on what you want and need, that your model could offer to the apps and AI you use? Why couldn’t they negotiate when they disagree, and let you make the final choice about when and how they can access you and your data?

When I say Your Model, Your Terms, I mean it literally. I think our personal models should also be run on our personal terms and conditions. I think this would help protect against a further consolidation of power by a few AI labs that are currently writing the rules of the future. I think it’s good for the market and good for society.

Intension was founded on the idea that technology should work for you. We’ve been building a consumer product that allows anyone to create a personal model from their data, on their devices, and put it to work. We’ve also been thinking about the other half of the equation: if it’s possible to have your own terms, what would need to be true for this to succeed, and how does this make the product we’re creating even more valuable.

A personal model is a working, compounding representation of you and your world: what you're doing, what you care about, who matters to you, how your body is doing, the patterns and signals that reveal what you cannot see, and the tools and systems you use to get it all done.

Companies already have versions of these models assembled from your behavior, tuned to their revenue and advertising goals. A model you own will be different in one critical way: it will be working only for you.

This goes beyond data storage or AI memory. A model isn’t yours just because it remembers more about you. What makes it yours is how it works, what it’s allowed to do, and the incentives that guide its decisions. Your model can evaluate and pick the best AI for what you're doing (a frontier model, an open one, a small one running on your phone), give it only the data needed, and keep it all working on your terms.

The most sensitive parts of that model should always sit closest to you, on your devices. Which brings us to the question of what exactly you could own and how it might work.

If you use AI today, it’s a bit like renting an apartment in someone else's building. You get to decorate it how you like, within the guidelines of your lease. While you probably wouldn’t hand your keys to strangers, there's usually a spare set behind the front desk that staff can use in special circumstances. In the best case nobody sees or hears what you do inside, but the hallway cameras still see exactly when you come and go. And at any time, they can sell the building to someone new, and you wouldn’t have any say.

The difference with a lease is you can still negotiate with the landlord, and they might compromise on the price you pay or the terms that govern what you can expect. When was the last time that happened with a tech product you used?

Now contrast that with owning a home. It’s yours, a space you own entirely. You can do what you want and no one else can change the rules on you. You don't ask permission to make a major renovation and the value of the work you put in is yours; you're building an asset you get to keep. Yes, you’re still part of a community; laws apply, the city sets codes, your neighbors have rights. But overall, you start off with better terms than in a rented space.

Taking this back to your model, you can think about your laptop, phone, and other smart devices as your home and the plot of land it sits on. That doesn’t mean you won’t want to rent an apartment or stay in a hotel sometimes, but the most important things should be in a space you fully control.

Today the experience is: you accept a product's terms, and then it gets access to you and your data in exchange for a service.

With a personal model you own, the process reverses: no app, agent, or AI gets access to you until it has accepted your terms.

Say an agent wants to plan your next trip. It asks your model for context to help it do its job. Your model would share your calendar constraints and preferences, withhold your health history, refuse training rights on your data, and cut off access when the trip is over, while keeping a record both sides can check. The agent accepts, or it negotiates with your model. If it won't agree to your terms, the job goes to one that will. It’s all based on a small set of choices you can understand, lawyers can enforce, and software can execute.

Some of the technical standards for this to be possible already exist. In 2025, the IEEE (the standards body behind Wi-Fi) published a standard nicknamed MyTerms (IEEE 7012-2025). It has all of the ingredients needed to make the experience I described possible. While it doesn't solve ownership completely, it proves we’re closer than we might think to a world where your terms can be real.

A personal model is a new kind of space: a digital one, made of data instead of brick. Analogies are helpful here, so let me explain.

A home you own is many things. It's the raw materials that make up the foundation, frame, and the furnishings that make it distinctly yours. It's the space you live your life in with others, where you can grow and create memories.

Similarly, a personal model is made of many different pieces. It's the combination of models, tools, and the accumulation of context that form a digital space that's uniquely yours.

Just like any other space you own, you could have the right to say how your model is built, who comes in and out, and what they can do while they're there.

No single body of law covers that space today, but the pieces exist. Property, contract, and a patchwork of privacy laws each have relevant concepts. But they need to be connected into one thing you can point to and call your own, in an experience that’s easy to adopt.

To see how we can establish new digital rights, we can look back in time. At the dawn of the internet, ordinary people could suddenly share their digital work with the whole world, but lacked a simple, standardized way to tell others how it could be used. Copyright laws weren’t helpful. They were primarily written for publishers and record labels, and defaulted to an "all rights reserved" approach: nobody could copy, share, or build on your work without asking you first, even if you'd happily say yes. The only way around that was to hire a lawyer and write your own license, which was not realistic for most people. Even then, there was no shared vocabulary that people, computers, and the legal system could all understand.

So, an organization called Creative Commons created a handful of free licenses and a common vocabulary that anyone could apply to their work. It took less than a minute to choose what rights you wanted to keep, and your terms were expressed in plain language you can read, in legal code the law can enforce, and in metadata a computer can understand. More than a billion photos, articles, and songs now carry a Creative Commons license (every page of Wikipedia, entire museum collections), each shared on terms its creator picked instead of terms a platform imposed.

Glenn Brown was the executive director of Creative Commons from 2002 to 2005 and one of its early architects. He's been working with us at Intension on this idea, exploring how ownership and digital rights could apply to personal models in this new era.

While we build our core product, I hope that sharing our early thoughts and ideas connects a few dots and sparks conversation, because there is so much to figure out and debate. The answers shouldn't come from us alone.

The defaults for how AI shapes our lives will get set once, and they're being set right now by a small number of people around the world. Everyone will live with these choices for decades. I think they should be discussed and decided in public, by anyone who cares about their freedom and personal power in the age of AI.

If that's something you care about, add your name here. You'll get updates, and an early invitation to create your own model, on your terms.