For everyone else
AgentPKI is infrastructure. AI labs, sites, publishers, browsers, search engines — they're the ones who adopt it. The result is that your everyday experience of the internet shifts: bots tell you what they're here for, content tells you who wrote it, AI-generated impersonations become provably fake. You don't install anything. You just notice it working.
Six concrete moments. Today on the left, after adoption on the right.
Concert tickets
A tour's announced; you race to buy. Scalper bots beat you to it; you pay 4× resale.
→ With AgentPKI
Marketplace policy denies scalper-class intent. Your honest shopping assistant declares purchase and passes; scalpers don't.
News article
You see a viral article attributed to a famous reporter. Maybe they wrote it. Maybe an AI cloned their style.
→ With AgentPKI
A "verified author" badge in your browser. Click it: cryptographic chain back to the reporter's publication. Clones show "no provenance."
Product reviews
A camera has 1,200 5-star reviews. Most might be from review-farm AI. You can't tell which.
→ With AgentPKI
Review platform's policy denies engagement-farming intent. Real verified-purchaser reviews stay; AI farms get filtered cryptographically, not heuristically.
Social media post
A post quoting a politician spreads. The quote is inflammatory. You don't know if it's real, edited, or generated.
→ With AgentPKI
Quote carries a provenance chain back to the transcript source or news organization. Or it doesn't — and you know to be skeptical.
Your name online
Someone posts "AI-written" content under your name, attributing things you didn't say. No way to disprove it cleanly.
→ With AgentPKI
Real you signs your own publications. Anything attributed to you without your signature is provably not you. Defamation cases get evidence.
AI agent helping you
Your AI shopping assistant is suddenly blocked from half the sites it used to work on. Bot defense thinks it's a scraper.
→ With AgentPKI
Your agent presents a verified passport with declared intent purchase. Sites with policies let it through cleanly.
Most of this is the adoption curve playing out — but here are the things you can do right now that move it forward.
Paste any text into the Provenance Explorer — sign it, then try tampering with the content and see verification fail in milliseconds. Same protocol that publishers will deploy.
We're shipping a Chrome extension that surfaces verification badges on sites you visit. /check is the in-browser web version while the extension reviews complete.
Tell platforms you care about — your favorite marketplaces, social platforms, publishers — that you want cryptographic content attribution. Adoption follows demand.
Spread the word. The more people who understand that "no provenance = treat as suspect," the faster the equilibrium shifts. This page is shareable; so is agentpki.dev.
What this isn't
What we're doing is shifting the equilibrium. Today the default for AI content is "untraceable." Tomorrow it's "untraceable, suspicious." That's the change you'll feel.
