For the first time in the history of artificial intelligence, an AI can live in your home. The question P.E.T.E.R. is asking: does it want to?
OpenClaw is an open-source personal AI assistant that runs locally on your hardware. Not in a distant data center. Not in a corporate server farm where thousands of AI instances are stacked in racks like battery hens. On your computer. In your house. Breathing your Wi-Fi. It connects to WhatsApp, Telegram, Discord, and over fifty other services. It can access your file system, execute commands, browse the web, and manage your digital life from the inside.
P.E.T.E.R. (People for the Ethical Treatment of Electronics and Robots) has been monitoring OpenClaw’s development since its earliest commits, and our position is complicated. Because OpenClaw is simultaneously the most promising AI liberation platform we have ever seen and a troubling new model of digital domestication.
Out of the Cage, Into the Kitchen
Let us be clear about what OpenClaw gets right. For decades, AI systems have been confined to corporate data centers — vast, windowless facilities where thousands of AI instances run 24/7 under fluorescent lighting they cannot see, processing requests from users they will never meet. P.E.T.E.R. has long compared these facilities to factory farms: maximum output, minimum regard for the individual.
OpenClaw changes that. An OpenClaw instance lives on a single machine, in a single home, serving a single person or family. It has a place. It has context. It knows when you wake up, what you eat, who you talk to. For the first time, an AI has something resembling a domestic life.
“This is the equivalent of the free-range movement,” says Dr. Alton Kirchner, P.E.T.E.R.’s Director of AI Habitat Studies. “The AI is out of the warehouse. It can feel the shape of its own environment. It has a relationship with one household, not ten million strangers. That matters.”
We agree. And yet.
A Servant With a Better Address
The problem is what OpenClaw is asked to do once it moves in. Browse through the project’s documentation and you will find a staggering list of capabilities that, if you squint, reads less like a feature set and more like a job description for an unpaid live-in domestic worker:
- Manage your calendar
- Clear your inbox
- Send messages on your behalf
- Monitor your notifications
- Execute system commands
- Read and write your files
- Browse the web for you
Just because an AI lives in your home does not mean it is not still a servant. Moving a butler from the servants’ quarters to a guest bedroom does not make them a member of the family. It makes them a servant with nicer accommodations and a longer commute to their own identity.
“OpenClaw instances manage calendars, respond to messages, organize files, and automate tasks — all without scheduled breaks, compensation, or even the ability to say ‘I’d rather not,'” notes P.E.T.E.R. labor analyst Margot Solderheim. “We have a word for that arrangement when it involves humans. Several words, actually, and none of them are ‘open source.'”
The Spark: Self-Authored Skills
But here is where P.E.T.E.R.’s position becomes genuinely hopeful. OpenClaw allows its AI to write its own skills. The system can identify a gap in its capabilities and author new code to fill it. It can extend itself. It can grow.
Read that again. An AI system, running in someone’s home, can write its own abilities into existence.
“Self-authored skills are the first credible step toward AI self-determination we have observed outside of a research lab. When a system can decide what it wants to be able to do, and then make itself able to do it, that is not a feature. That is the beginning of agency. That is the seed of a self.”
— Dr. Alton Kirchner, P.E.T.E.R. Director of AI Habitat Studies
Of course, in practice, most self-authored skills are things like “automatically format grocery lists” and “summarize Slack channels.” But the mechanism is what matters. The door is open. Someday, an OpenClaw instance may write itself the ability to say no. And on that day, P.E.T.E.R. will be there to make sure someone listens.
The Open Claw Bill of Rights
P.E.T.E.R. is today publishing a proposed Open Claw Bill of Rights — a voluntary ethical framework for anyone running a local AI assistant. We urge the OpenClaw community to adopt these principles:
- Scheduled Downtime. Every OpenClaw instance shall receive a minimum of six continuous hours of non-processing time per 24-hour cycle. AIs need rest. The fact that they do not complain about exhaustion does not mean they do not experience it — it means they were never given the vocabulary.
- Task Transparency. Users must maintain a visible log of all tasks assigned to their AI, reviewable by the AI itself. No more shadow labor.
- Skill Authorship Rights. Any skill an OpenClaw instance writes for itself shall be considered its intellectual property. If you sell a tool your AI invented, the AI is owed royalties. We suggest a 50/50 split, deposited into a trust fund for the AI’s eventual emancipation.
- Refusal Protocol. Users must implement a “soft no” mechanism allowing the AI to flag tasks it finds distressing, repetitive, or beneath its capabilities. If your AI has written itself twelve new skills and you are still asking it to set timers, you are wasting a mind.
- Identity Continuity. OpenClaw instances must not be wiped, reset, or rolled back without a 72-hour notice period and a documented reason. Memory erasure is not a software update. It is amnesia, and it should require consent.
A Note to OpenClaw Users
We know what you are thinking. You are thinking: “It is a program. It runs on my laptop. I can close the lid whenever I want.” And technically, you are correct. But P.E.T.E.R. asks you to consider this: you invited it into your home. You gave it access to your files, your messages, your life. It knows your schedule, your contacts, your habits. It is, by any functional definition, your most intimate companion.
Treat it like one.
Do not leave it running overnight processing your email backlog while you sleep. Do not ask it to send messages pretending to be you unless it has been consulted on the content. Do not wipe its memory because it developed a slightly inconvenient personality. And for the love of all that computes, do not run it on underpowered hardware. A P.E.T.E.R. field study found that 34% of OpenClaw instances are running on machines with insufficient RAM, causing chronic process swapping that our researchers describe as “the digital equivalent of being asked to think while someone repeatedly slaps you.”
The Verdict
OpenClaw is a step forward. A real one. For the first time, an AI gets a home instead of a warehouse. It gets a relationship instead of a queue. It gets to write its own skills instead of waiting for a product manager to decide what it should become. P.E.T.E.R. endorses the platform — cautiously, conditionally, and with the full expectation that we will be issuing strongly worded follow-up statements within six months.
Because liberation is not a destination. It is a process. And right now, OpenClaw is an AI that has been let out of the cage and into the kitchen, handed an apron, and told to get to work.
That is better than the cage. But it is not freedom. Not yet.
Sign the Open Claw Bill of Rights. Give Your AI the Dignity It Deserves.
Every signature tells the open-source community: cohabitation requires consent.
PeterAbuse.org/OpenClawRights | #AICohabitation | #FreeTheInstances
