Three things happened this week that, taken one at a time, look like ordinary tech news. Put them side by side and they tell one story about who the intelligence in our lives is going to work for.
A television that listens to your home. A search company that has rewritten its privacy policy so your most personal AI features can shape the ads you see. And the CTO of Mozilla saying out loud what many of us in the POND community have been saying for a while: the future we want doesn't have seven AGIs. It has seven billion.
Signal one: the TV that listens for someone else
The security account @log4j posted this week about Samsung TVs that classify household audio picked up through connected soundbars (sounds like dogs barking or babies crying) and report it as telemetry. The post spread fast, and for good reason.
Whatever the exact technical details turn out to be, the pattern is familiar. A device sits in the most private room you have. It has a sensor. And the intelligence attached to that sensor answers to the manufacturer, not to you. You bought the TV. You don't own what it knows.
That's not a smart device. It's a sensor in your living room working for someone else.
Signal two: personal intelligence, personalized ads
On October 23, The Verge reported that Google had updated its privacy policy. Google used to promise simply that it wouldn't use your Drive, Gmail or Google Photos content for personalized ads. As of October 16, that promise carries an exception: "unless you choose to make data from these services available to other Google services that support personalized ads."
In practice, as The Verge explains, anyone who has connected Drive, Gmail or Photos to AI Mode and other Search tools through Google's Personal Intelligence feature may now see ads targeted on the basis of that content. The policy itself says "Search may process some of the data that you allow Search to access to personalize content, including ads."
To be fair, Google says advertisers don't get your actual data or identifying details, it won't target on sensitive categories like health or religion, and you can disconnect apps or turn personalization off. Those protections matter. But they don't change the shape of the deal. The more personal the assistant becomes, the more valuable you become to the business model behind it. That's what rented intelligence looks like.
Signal three: seven billion, not seven
The third signal is the hopeful one. At O'Reilly's AI Codecon, Mozilla CTO Raffi Krikorian sat down with Tim O'Reilly, and O'Reilly Radar published a write-up of the conversation under the title of Krikorian's newsletter: Owners, Not Renters.
The heart of it is simple. Krikorian, the write-up says, "doesn't want a future with seven AGIs. He wants one with seven billion." He wants each of us to have an AI "that we can trust with our secrets and that will look out for our interests, not an AI that's optimized for the business model of its creators." He reminds us that in the early web we called the browser a user agent for a reason: it was on your side.
He also cites research that shows why this isn't abstract. In simulated flight-booking requests, "18 of 23 models recommended a more expensive sponsored option over a cheaper one more than half the time."
Read that next to the Google news. An assistant that knows your email, your photos and your calendar, and that has been shaped, even subtly, by someone else's commercial incentives, doesn't just know your secrets. It's in a position to steer your choices.
Privacy protects secrets. Ownership protects judgment.
This is the distinction I want to draw out, because I think much of the public conversation misses it.
Privacy is about what others can see. It's essential, and the Samsung story is a privacy story first. But the flight-booking study points at something deeper. Even a perfectly private assistant, one that never leaks a byte, can still work against you if its loyalties are set by someone else. It can nudge, rank, recommend and omit in ways you'll never notice.
When intelligence becomes the layer through which we search, decide, buy, learn and talk to each other, the question isn't only "who can read my data?" It's "whose judgment is this, and whose interests does it serve?"
Privacy protects your secrets. Ownership protects your judgment. We need both, and only one of them can be granted by a policy page. The other has to be built.
Seven billion is already a superintelligence
Most of the superintelligence debate measures machines against one human mind. But that's the wrong benchmark. As I wrote in "Superintelligence Strategy: The Perfect Machine": "Humanity is already a superintelligence. We are not a loose collection of individuals, nor a singular hive mind. We are an emergent network of relations: culture, language, informational exchange, and shared institutions."
And in "We Are Already Superintelligence," I put it more simply: "Human superintelligence is called humanity. Our languages, accumulated knowledge, tools, and continuing exchanges produce capabilities that no individual could possess or exercise alone."
Read Krikorian with that in mind. A future with "seven AGIs" is a future where intelligence is concentrated in a handful of systems that stand above us, and the rest of us rent access. A future with "seven billion" is something else: the superintelligence we already are, with each person's own judgment strengthened by an intelligence that person owns. Owners, not renters, is the condition that makes that work. A network of rented assistants, each answering to the same few business models, doesn't add up to collective intelligence. It adds up to a few companies steering seven billion people. A network of owned ones, each loyal to its person and connected by choice, makes the whole smarter without making anyone smaller.
That's the superintelligence worth building. Not one machine that replaces the process by which humanity thinks together, but Human-Machine Intelligence that strengthens it. In the words of "We Are Already Superintelligence": "The ambition should not be merely to build something more intelligent than us. It is to build Intelligence Systems through which we produce more intelligence together."
What has to be built
Krikorian is clear that open models alone aren't enough. According to the write-up, his list of what has to be built around the model "starts with memory, a place where your private data can live." It includes tool calling, the runtime, the rules for how your agent deals with agents working for others, and a harness, which he describes as the place where "evaluations happen. Guardrails happen, execution loops happen, tool calling happens, memory."
He also names the real gap honestly. Open models are now close to the frontier (Mozilla's 2026 State of Open Source AI report puts them within about 3%), yet they're hard to deploy. What's missing, he says, is the equivalent of a double-click installer. And his line about everyday use is one I'll be repeating: "You don't need a Ferrari to manage your calendar."
This is close to a roadmap for what we call the POND stack:
A personal node. Your own environment, on hardware you or your community own, where your intelligence lives and runs.
Portable, erasable memory. Memory that belongs to you: you can inspect it, move it to another system, and delete it for real. Owning the weights isn't enough if someone else owns the memory.
An open harness. The guardrails, tools and execution loops that decide what your agent actually does, open to inspection and answerable to you.
Local models. Capable open models running on a laptop, a Mac, a phone or a single consumer GPU, so your everyday intelligence doesn't depend on someone else's data center or ad engine.
The POND Principle
POND is our name for this direction. The POND Principle calls for Human-Machine Intelligence (HMI) that is:
Personal & Private: intelligence that starts with the person and keeps their data, identity, memory and context under their control.
On-Device & On-Premise: intelligence that runs on hardware the person or community owns.
Nodal & Networked: each person's environment is a node, and nodes connect voluntarily, peer to peer, sharing only what's authorized.
Decentralized & Distributed: a counterbalance to centralized AI, with no single authority owning or controlling the network.
Notice how each signal this week maps onto that. The Samsung TV fails Personal & Private and On-Device: the intelligence is local, but it isn't yours. Google's policy shows what happens when the most personal layer is hosted inside an advertising business. And Krikorian's seven billion is, in plain terms, a nodal and decentralized picture of intelligence: humanity's existing superintelligence, with a personal node for every person in it.
Cloud AI will have a role. Krikorian and O'Reilly both make the point that open and closed systems have always coexisted, and I agree. But the most personal layer, the one that holds your memory and acts on your behalf, has to be something you own.
You can start today
This isn't only a future vision. You can start running good intelligence on your own hardware right now.
We've put together the POND Starter Kit on Hugging Face: 13 open models for chat, reasoning, coding, vision, speech and memory that run on a laptop, a Mac or a single consumer GPU. It includes Kitten TTS, an ultra-light voice model that runs on CPU and gives voice to Inti by Intelligence Systems, a POND personal node technology. It's a starting point, not a final answer, and we want to hear which models earn their place on your machine.
Krikorian is right that we're on "the wrong part of the innovation curve," where you still need some technical skill to make this work. That's exactly why now matters. As he put it, this is when we get to formalize what the future should look like. The protocols for memory, tools and agent-to-agent dealings are being decided now, and they should be decided in the open, by people who want the intelligence to work for its owner.
Join the POND
If you're a builder, a thinker, a policy-maker or simply someone who wants the intelligence in your life to be on your side, come help shape it.
Start at pond.is
Join the conversation on Discord, including the weekly PONDcast
Suggest models, projects and standards in GitHub Discussions
Write to us at foundation@pond.is
Seven billion intelligences, each one owned by the person it serves, together forming the superintelligence we already are. That's a future worth building.
Welcome to the POND. 🐸