On-device content safety

Take the hate out of your feed.

valyou quietly hides hateful, violent, and rage-baiting posts on Facebook, Instagram, and X — before you ever see them.

Runs on-device AES-256 encrypted Zero network No account
your feed

The feed isn't neutral. It's tuned to keep you angry — because outrage is what holds attention. valyou tunes it back.


How it works

A classifier that lives on your device.

Every post is read, scored, and acted on in the moment — locally, in memory, and then discarded. No cloud call, no external AI service, nothing to subscribe your data to.

READ

Understand the post

valyou pulls the real text of each post, comment, ad, and message — anchored to the page's accessibility structure, so it keeps working as sites reshuffle their markup.

extractors · ARIA-anchored
Facebook · Instagram · X

SCORE

Weigh the signals

Structural detection and an on-device machine-learning model combine into one calibrated score, with damping for news, quotation, and counter-speech so context isn't punished.

structural + ML (fastText-family)
noisy-OR combination

ACT

Hide, blur, or tag

Cross a threshold and the post is collapsed, blurred, or tagged — your choice per category. A calm control lets you reveal anything with a tap.

per-category actions
reversible · your call

What it catches

Five things you didn't ask to see.

Turn each category on or off, and decide what happens when it's found. These are the defaults.

Racial hate

Hide

Slurs, dehumanizing language, and coded racist tropes — including obfuscated and “dog-whistle” variants.

Gendered & sexist hate

Hide

Misogynistic and gender-based attacks, harassment framed as “just an opinion.”

Violence & threats

Hide

Calls for harm, glorified violence, and language that targets people or groups.

Harassment

Blur

Pile-ons, demeaning insults, and content aimed at wearing a person down.

Rage bait

Blur

Posts engineered purely to provoke — the outrage machine's raw material.

Every video

Tap to play

Nothing autoplays. By default each video stays hidden until you choose to play it — you decide what earns your attention.

Private by design

Nothing you browse ever leaves your device.

valyou has no servers and no accounts. The classifier, the model, and the encryption all live on your phone or in your browser. It can't see your feed — because your feed never reaches it.

That's also why it works without any paid AI service behind it: the intelligence is bundled in, not rented.

No data collectedNo tracking, no analytics, no ads. There is nothing to leak.
Encrypted at restYour settings sit in the OS keystore, sealed with AES-256-GCM.
Zero network accessThe engine makes no outbound request of its own — ever.
Yours to inspectApache-2.0 licensed. The way it decides is open to read.

Everywhere you scroll

One filter. Every screen.

The same engine — the same decisions — in your browser and in your pocket.

Chrome extension

Manifest V3. Filters the desktop web the instant a post loads, with a popup and full options page.

● Core product

iOS app

A private, content-filtering browser for Facebook, Instagram, and X — the same engine, injected into every page.

● Built · store submission next

Android app

The same filtering browser, packaged for Play — no accessibility hooks, no overlays, just a calmer feed.

● Built · store submission next

Store listing

Ready to publish.

The exact copy for the Chrome Web Store — clear title, a summary under 132 characters, and a description that leads with the features people install for.

Title
valyou — Block hate, violence & rage bait
Summary
Hides hateful, violent, and rage-baiting posts on Facebook, Instagram, and X — entirely on your device.
108 / 132 characters
Description
valyou is a content filter for the social web. It reads each post as it loads and quietly hides, blurs, or tags anything hateful, violent, harassing, or engineered to make you angry — so your feed stays yours. Everything runs on your device: no servers, no accounts, and nothing you browse is ever uploaded.
  • Filters Facebook, Instagram, and X — feed, comments, ads, and messages
  • Catches racial and gendered hate, violence and threats, harassment, and rage bait
  • Hide, blur, or tag — you choose the action for each category
  • Blocks autoplay and hides every video until you tap to play
  • On-device machine-learning model — no external AI service required
  • Encrypted settings, zero network access, and no tracking of any kind
Category
Social & Communication · Works with a subscription; the engine needs no paid backend.