Introduction

IMAGXP Protocol: The Official Standard (v1.2)

Real. Verified. Universal.

The current web model is broken.

  • The Problem: AI Agents aggressively scrape data, often spoofing user agents to bypass basic protections.
  • The Reaction: Publishers block aggressively using IP bans, captchas, and login walls.
  • The Result: A "Dark Forest" Internet where useful data is locked away and creators aren't compensated.

IMAGXP is not a product; it's a protocol. It replaces manual "Allowlists" with Automated Identity.

⚠️ Important: We are a Protocol, Not a Firewall

Can IMAGXP physically stop a "Trillion Dollar Scraper"? No.

  • The Firewall's Job: Companies like Cloudflare stop attackers who use residential proxies ("Steve from Ohio") to look like humans. That is a physics problem.
  • Our Job: We solve the Legal Problem.
    • Currently, Scrapers use "Plausible Deniability" ("We didn't know you didn't want us").
    • IMAGXP removes that excuse. By serving a cryptographically signed "Payment Required" signal, any bypass attempt becomes Intentional Hacking (CFAA Violation).
    • The Result: Public companies will not risk criminal liability. They will negotiate.

IMAGXP acts as the "Digital Treaty". Cloudflare acts as the "Wall". You need both.

The Core Mechanism: "Connect, Verify, Decide"

Instead of guessing if a visitor is a bot, IMAGXP automates the handshake:

  1. AI Agents sign every request with a cryptographic key (ECDSA).
  2. Publishers automatically verify this signature against the Agent's public DNS record ("Digital Passport").
  3. Strict Rules are applied after identity is proven (e.g., "Pay $0.001", "No Training", "Show Ads").

How it works "In the Wild" (The Automatic Flow)

Scenario: OpenAI Bot visits "New York Times" (IMAGXP-Enabled)

Step A: Discovery (The Handshake)

  • Action: The OpenAI Bot tries to visit nytimes.com.
  • Detection: The Bot sees a Signal (in robots.txt or a Header) saying: "I speak IMAGXP."
  • Result: The Bot switches from "Scraper Mode" to "Polite IMAGXP Mode" and prepares its Digital Passport.

Step B: Identity (The Passport Check)

  • Action: The Bot sends a request signed with its Private Key.
  • Publisher Check: The Publisher asks: "Is this REALLY OpenAI, or a hacker?"
  • Verification: The Publisher checks openai.com/.well-known/imagxp-agent.json.
    • Success: The Key matches. This IS OpenAI.
    • Fail: The Key doesn't match. It's a Fake. BLOCK.

Step C: The Rules

Now that we know it's OpenAI, we check the Publisher's Terms (Payment, Training, Ads).

The 2026 Context: Why Now?

The "Free Scraping" era is ending. The "Ad-Supported" era is beginning.

As of 2026, every major AI lab is pivoting to an Ad-Supported Model to cover the massive costs of inference ($100B+ infrastructure).

  • OpenAI: Officially testing ads in ChatGPT (2026 rollout) to reach a projected $25B in ad revenue.
  • Google: "AI Overviews" ads are expanding globally, fundamentally changing Search traffic.
  • AWS: Building "Generative Ad" infrastructure (Amazon Bedrock) for global marketers.

The Risk for Publishers: If you do nothing, your content will be used to generate answers that show ads for other products, while you get $0.

The IMAGXP Solution: IMAGXP allows you to participate in this new economy.

  • Proof of View: You can demand that the Agent proves it showed your ad (or gave you attribution) before you give it the data.
  • Revenue Share: You can block Agents that refuse to share the ad revenue.

We are the negotiation layer for the $25B AI Ad Economy.

The Vision

We don't need to break the open web to save it. We just need a standard way for machines to say "Hello" and "Here is my ID". IMAGXP is that standard.