Trust Hound

Your feed's loyal watchdog. Flags false and misleading claims in posts and videos as you scroll, with sources for every verdict.

Trust Hound reads along while you scroll the major social media sites and short-video feeds. When a post or video makes a claim that doesn't hold up, it puts a small badge on it, with the sources so you can see for yourself. WHAT THE BADGES MEAN ❌ False: reliable sources clearly say otherwise ⚠️ Misleading: partly true, but missing context or exaggerated 🟡 Mixed: some claims hold up, some don't ✅ Checks out: reliable sources back it up ❓ Can't verify: too new or too little evidence to say ⚠️ Unproven claim: the only support is the seller's own marketing or numbers 🔍 Claim found: a lower-stakes claim you can tap to check 👤 Personal claim: something only the poster can prove, like "I make $1,000 a week," with a note when they're selling something 🐕 Satire: the post links to a known satire site, so it's a joke, not a lie Tap any badge for a plain-English explanation and links to every source. WHAT IT CHECKS Out of the box, claims about health, money, scams and safety are checked automatically, because that's where people get hurt or lose money. Other claims get a badge you can tap if you're curious. You can change the topics any time. IT CHECKS THE ADS, TOO Sponsored posts and video ads are always checked and tagged "Ad". Trust Hound also points out how the ad is selling to you: fake urgency, testimonials nobody can verify, celebrity endorsements that don't check out, or the company's own numbers. It makes ads obvious, even when the platform doesn't. Social sites keep shrinking their ad labels into tiny gray text. Trust Hound puts a clear "Ad" tag right on the badge, and a "Selling" tag on posts that are quietly pushing a product, course or "link in bio" with no ad label at all. IT FOLLOWS THE LINK A post can say one thing while the article it links to says another. When a post shares a link, Trust Hound opens the article and checks whether the post describes it honestly. Links to satire sites get a "Satire" badge instead of a verdict, and links to sites fact-checkers have repeatedly caught publishing hoaxes get a heads-up. It reads pictures, too: screenshots of headlines, memes with text, charts. It works on videos too. Trust Hound reads captions and the words shown on screen, and listens to what's said using a speech model that runs on your computer, so the audio never leaves it. Every verdict shows what it heard. WHAT IT LEAVES ALONE • Politics: posts about elections, politicians, parties and hot-button issues are skipped by default • Opinions, jokes, memes and satire • Comments and replies (it checks the post itself), friends' Stories, items people list for sale, and search results • Your privacy: no account, no tracking, no ads HOW YOU PAY Trust Hound runs on Claude, an AI from Anthropic. You connect your own Claude account and pay Anthropic directly for the checks you use, with nothing added on top. For most people that's a few cents to about a dollar a day. A daily spending cap means it can never run up a surprise bill, and each post is only checked once. SET UP IN ONE MINUTE 1. Install Trust Hound. 2. Keep the suggested topics or pick your own. 3. Paste your Claude API key (the welcome page shows you how to get one). 4. Go back to scrolling. A badge is a nudge to look closer, not the final word. AI can make mistakes, so every verdict shows its sources. Questions? support@trusthound.dog Privacy policy: https://trusthound.dog/privacy.html