There is a paragraph in the middle of Mark Zuckerberg’s new essay on superintelligence where he describes baking with his young daughter. His AI agent plans the recipes. It orders the ingredients. It offers suggestions while the two of them work.
He offers this as evidence that the future is already good, and it is the most revealing thing in six thousand words. He has looked at an afternoon with his kid, identified every part that requires deciding, and assigned all of it to software. What survives is the part that photographs well.
It reads like a parody because it is an undistorted picture of a real ideology. Not Zuckerberg’s personal quirk. The ideology. The same one running through every keynote and every investor letter from the same handful of companies: that friction is the problem, that human judgment is the bottleneck, and that the correct number of unaided decisions in a day is zero.
The essay is called “The Future Is For Everyone.” I do not want the future it describes. Not a slower version, not a regulated version, not a version with a better privacy policy. And I do not accept the premise that the people building it are benevolent and acting in good faith.
The hardware comes wrapped in privacy assurances, and every one of them is addressed to the wearer. Your data, your encryption, your control. The person across from you on the sidewalk gets nothing. No screen, no toggle, no question. That is not an oversight to be patched in version three. A camera on a face is only useful because it is pointed at other people, and the entire value proposition depends on those people not being consulted. The product requires our non-consent.
This is not new behavior dressed up in new hardware. Facebook was built on non-consent. Sharing data with third parties. Sharing friend and network information. Friends of friends. Advertiser data. The entire foundation of that company rests on the fact that it was built without informed consent and then kept running long after everyone understood what that meant.
And Meta is only the most documented case. Build first and ask for forgiveness never is how tech companies operate. This attitude built the social web, and it is now being used to justify feeding mass domestic surveillance and training generative AI on the creative and intellectual output of billions of people who were never asked and never compensated.
Zuckerberg’s essay includes a passage about personal superintelligence protecting individuals from government tyranny. Rights by default, restrictions only where necessary. The threat under examination is always the state. Never the intermediary. Never the company that issues the tool, prices the tool, hosts the tool, writes the policy governing the tool, and revises that policy whenever it likes.
This framing has become so routine in Silicon Valley that it barely registers as an argument anymore. It is closer to a liturgy. Government is the danger; technology is the remedy; the company selling you the remedy is not a power center, just a surveillance service provider. It is a prosperity gospel for the secular class: believe in the platform, and abundance will follow. Give your data, your attention, your consent, and you will receive connection, convenience, liberation. The fact that the abundance keeps accruing to the preacher is not supposed to trouble you.
I are not speculating about how Meta behaves when it holds power. By their fruits you will know them, and the fruits are in the company’s own documents. Internal research on what Facebook’s products did to teenage girls, which the company suppressed until a whistleblower carried it out the door. A human rights assessment the company itself commissioned, finding it had failed to prevent its platform from being used to incite violence in Myanmar. Cambridge Analytica.
Zuckerberg published his essay about the “glorious” future in the same week that Meta was on trial in federal court in Oakland, sued by a bipartisan coalition of 51 state attorneys general for designing Instagram and Facebook to be addictive to children and then lying about it. The states’ complaint alleged that Meta used its platforms to “entice, engage, and ultimately ensnare youth and teens,” that it knew about the harm its products caused to the mental and physical health of young users, and that it deliberately misled the public about the severity of those harms. A former Meta engineer testified that the company’s own internal studies showed young users were exposed to harmful experiences at far higher rates than Meta ever publicly acknowledged.
Meta settled midway through the trial’s second week. The proposed settlement requires Meta to pay up to seventeen billion dollars over ten years, institute default daily time limits and nighttime blocks for users under eighteen, ban the display of like counts to minors, ban cosmetic surgery filters for minors, offer a non-algorithmic feed option, and submit to an independent auditor with broad access and the authority to report directly to the attorneys general. The case has been compared to the tobacco litigation of the 1990s. That comparison is apt, because the core allegation is the same: the company knew its product was harmful, it had the data proving it, and it chose to keep selling.
Seventeen billion dollars sounds large until you consider that Meta reported roughly sixteen billion in net profit during their 2026 Q1 earnings call. Spread over ten years, the settlement amounts to about one week’s earnings per year. The real cost to Meta is not the money. It is the public opinion of the company that now has a federal consent judgment in which the company agreed, under oath, to stop doing things it had spent years insisting it was not doing. The injunctive terms read like an inventory of design choices the company made on purpose: addictive feeds, hidden metrics, suppressed safety research, ignored age restrictions. Every safeguard in the settlement is a safeguard Meta chose not to build when it could have.
This is the record of the person who is now explaining that he should be trusted with a general-purpose intelligence in everyone’s pocket and a camera on everyone’s face. Meta has not earned the benefit of the doubt. They have spent it.
The case against these companies has never been that they are annoying. It is that they are corrosive, and that the people running them have told us in plain language what they are building next. So I am leaving, and I know exactly how little this accomplishes.
I have spent months working to de-Google my life, a subject for another post, and now I intend to de-Meta my life. We have been told what the future looks like, and I am opting out of that future.
To be clear, I am not against AI. I use AI every day. It helps me research, synthesize, proofread, debug, draft, and think. It is a genuine tool in my work and I have no intention of giving it up. What I am against is the Silicon Valley ideology that has attached itself to the technology. An ideology that treats human judgment as an inefficiency to be optimized away, human attention as a commodity to be extracted, and human consent as a cost to be engineered around. The ideology that looks at a person and sees a bottleneck. That has never been a technology problem. It is a values problem, and it predates AI by a decade. Social media was built on it. The surveillance economy was built on it. AI is just the latest product to carry it forward, and Zuckerberg is just the latest executive to write six thousand words dressing it up as progress.
He described a scene in which the only part left for the humans was the manual labor. The exact work we were told would be automated by robots, so that we could have more time for the creative and intellectual work that AI is now replacing as well. First they said the machines would do the drudgery so we could do the thinking. Now they say the machines will do the thinking so we can do the drudgery. At no point in this story does the person end up with more.
That is the future on offer. The only decision it leaves me is whether I am in it.
I am not.