🤚 The Open-Palm Disclosure

In a move that combined the subtlety of a freight train with the transparency of frosted glass, Google has confirmed that it is now collecting and storing photos, files, audio recordings, and video from users across its product ecosystem — and feeding them directly into its AI training pipeline. Every user is opted in by default. The announcement arrived via a quiet email in June 2026, the corporate equivalent of mumbling bad news into your collar while leaving the room.

The affected services read like a tour of your entire digital life:

• Google Search — your queries, now curriculum
• Google Lens — every photo you pointed your camera at to identify a plant or a suspicious mole
• Search Live — your voice searches, preserved for posterity and gradient descent
• Google Maps, Shopping, Flights, Hotels — your travel plans, purchase intent, and that time you searched for “cheapest hotel near divorce lawyer”
• Google Translate — including practice audio, because even your attempt at conversational Portuguese deserves to train a foundation model
• Google News — what you read, how long you read it, and presumably what you rage-clicked

Google’s official statement was a masterpiece of passive-voice absolution: “Like your Search Services History, your saved media is also used to develop and improve Google services and technologies, including AI models and safety measures.” The word “also” is doing more work in that sentence than most of Google’s ethics board.

👐 The Two-Handed Consent Charade

Now, to be fair — and fairness is something we dispense here with great reluctance — Google did provide an opt-out mechanism. Two of them, in fact. On two separate pages. That you must find independently. Like a scavenger hunt designed by a company that earns $300 billion annually from knowing things about you.

The first page is the Search Services History settings, where you must uncheck a “Save Media” box that is separate from your search history toggle. The second is the Search Services Personalization page, which controls a different but overlapping set of preferences. You can also configure automatic deletion of your data after 3, 18, or 36 months — a range that suggests Google’s product team asked “how long should we keep people’s voice recordings?” and three different executives answered simultaneously.

The architecture of this opt-out is not accidental. It is the privacy equivalent of hiding the emergency exit behind a vending machine. The default is enrollment. The reversal requires navigation. And the average user — the one who still has “password123” on at least two accounts — will never touch either page.

🌿 The Gentle Awakening

It is worth pausing to appreciate the philosophical arc of Google’s relationship with your data. In 2004, the company’s motto was “Don’t Be Evil.” In 2015, they restructured under Alphabet and quietly retired the phrase. In 2026, they are storing your voice recordings to train AI models and the opt-out requires visiting two separate URLs.

This is not a slippery slope. This is a waterslide, and you were placed on it at birth when your parents created a Gmail account to share baby photos.

Google is not alone in this, of course. Meta has been training its AI on users’ photos, camera roll content, and video captured by Ray-Ban AI glasses — a device that combines fashion eyewear with surveillance capitalism in a package that somehow costs less than a pair of Warby Parkers. The entire industry has collectively decided that the data you generate while using their free products is, in fact, the product. You knew this. They knew you knew. They are now simply being marginally more honest about it, which in Silicon Valley counts as radical transparency.

👑 The Gold-Leaf Reckoning

The deeper question is not whether Google should train on your data — they were going to regardless, and the terms of service have always been written by people who bill at $1,200 an hour — but whether the opt-out theater matters at all. The data Google has already collected from two decades of your searches, emails, locations, and late-night YouTube spirals is already baked into models that are already deployed. Opting out now is like closing the barn door after the horse has been tokenized, embedded, and fine-tuned into a conversational interface.

The EU’s GDPR regulators will presumably have opinions. The FTC will presumably issue a statement expressing concern. Congress will presumably hold a hearing where a senator asks a Google executive to explain what an AI model is, and the executive will answer in a way that is technically correct and spiritually meaningless. The cycle continues.

Meanwhile, two billion users just became unpaid contributors to the most expensive training dataset ever assembled. Their compensation is the continued privilege of using Google Search for free — a service that now returns AI-generated summaries that are occasionally wrong and always confident, much like the company itself.

“The opt-out button is real, functional, and located on two separate pages behind three menus and a sense of existential resignation. We consider this best-in-class transparency.” — The Slap of Wisdom Privacy Compliance Division, training this article on itself as we publish it