How will publishers and agents pay each other?
If agents are the new first reader, the open question is the meter. How does a publisher get paid when a model — not a human — consumes the page? And what does an agent pay for that’s worth more than a free crawl?
Free crawl, paid answer. The publisher gets neither.
Ad-supported publishing assumed a human eyeball at the end of the funnel. Agents break that assumption — the answer is delivered without the page being seen. Three things are happening in parallel: infrastructure providers are bolting tollbooths onto crawl traffic, a handful of model labs are signing direct licensing deals with large publishers, and a long tail of independent sites are getting scraped for free with nothing in return. The protocols below are the early attempts to put a meter on the wire — most lean on HTTP 402 Payment Required, a status code that sat dormant for thirty years until agents needed it.
Cloudflare Pay-Per-Crawl+ open
Cloudflare-hosted sites can require AI crawlers to pay a publisher-set price per request before being served. Settled via Cloudflare; uses HTTP 402 + signed bot identity. Announced July 2025; now part of AI Crawl Control.
Publisher toggles it on in the Cloudflare dashboard, sets a price (Allow / Charge / Block), picks which bots get charged. Cloudflare is Merchant of Record.
Bot hits the URL, gets 402 Payment Required with terms, retries with payment intent.
Tollbit+ open
Third-party paywall-for-bots. Publisher integrates Tollbit; AI crawlers register, get rate cards, pay per fetch. Acts as clearing house. Distributed via partners including Fastly, Akamai, and Arc XP.
Drop-in script or edge integration; dashboard for pricing and analytics. Two licence types: summarization and full-display.
Agent authenticates against Tollbit, fetches under metered terms.
ProRata / Gist Answers+ open
Attribution-based payout: when Gist Answers cites a publisher, the publisher gets a share of the query/ad revenue, weighted by contribution. ProRata raised a $40M Series B in September 2025.
Publisher opts in, gets attribution receipts and payouts. Split is 50/50 between ProRata and the publisher pool.
Model logs which sources contributed to each answer; clearing house computes splits.
Direct licensing (OpenAI, Anthropic, Microsoft, Amazon)+ open
Bilateral contracts between a model lab and a named publisher. Flat fee or revenue share for training and/or retrieval rights. Examples: AP–OpenAI (2023), Axel Springer–OpenAI, FT–OpenAI, News Corp–OpenAI, Condé Nast–OpenAI, Hearst–OpenAI, NYT–Amazon, NYT–Anthropic settlement (Dec 2025).
Press release; opaque terms. Some deals require source-card citations (e.g. NYT–Anthropic).
The lab’s crawler is whitelisted; content may be fed via private feed.
RSL — Really Simple Licensing+ open
Open standard for declaring machine-readable licensing terms in robots.txt and Schema.org / RSS — supporting attribution, pay-per-crawl, and pay-per-inference. Launched 10 September 2025 by the RSL Collective, co-founded by RSS co-creator Eckart Walther and former Ask.com CEO Doug Leeds. Spec at rslstandard.org/rsl.
Publisher publishes a license file; standard schema integrates with existing robots.txt.
Agent fetches the license, confirms compliance, transacts (or doesn’t).
x402 — HTTP 402 for agents+ open
Open payment protocol over HTTP, launched by Coinbase in 2025 and now governed by the x402 Foundation (co-founded with Cloudflare, September 2025). Stablecoin settlement on Base, Solana, Polygon, Arbitrum, World. Open source on GitHub.
Invisible — pure machine plumbing.
Agent hits resource → server returns 402 with PAYMENT-REQUIRED header → agent returns PAYMENT-SIGNATURE → facilitator settles → resource returned.
Microsoft Publisher Content Marketplace+ open
Publisher Content Marketplace (PCM) is a marketplace where publishers can license content to be sold to agents, but this is still a pilot for now. Announced Feb 2026
Publisher lists content with terms; Microsoft routes payment.
Copilot products consume licensed feeds.
Affiliate / embedded commerce in agent answers+ open
Not a protocol so much as a business model: the agent’s answer includes purchase links that pay the publisher (or the agent vendor) on conversion. Perplexity, Arc Search, and others have shipped variants; Perplexity also runs a publisher revenue share with Time, Der Spiegel, Fortune, Entrepreneur, and The Texas Tribune.
Click-through from agent answer to checkout.
Agent resolves intent → product → affiliate-tagged URL.
What does NOT work yet+ open
robots.txt as a payment signal (no payment, just consent — see the RSL extension which adds the missing meter), Creative-Commons-style “please pay” badges, ad injection inside model outputs (rejected by labs so far), and any scheme that requires every crawler to voluntarily honour terms with no enforcement layer.
- One protocol crosses ~50% of top-1k publishers without a single-vendor chokepoint.
- A model lab publishes per-publisher payout data verifiable by the publisher.
- Independent publishers report meaningful revenue (>5% of total) from agent traffic.
- x402 or RSL gets crawler-side adoption from at least two of the top three labs (OpenAI, Anthropic, Google).
agent manifest (json)
{
"@context": "https://opensourceseo.org/agents/v0",
"@type": "StackQuestion",
"layer": "economic",
"question": "money-flows",
"title": "How will publishers and agents pay each other?",
"prompt": "If agents are the new first reader, the open question is the meter. How does a publisher get paid when a model — not a human — consumes the page? And what does an agent pay for that's worth more than a free crawl?",
"sections": [
{
"slug": "hook",
"label": "Hook",
"kind": "hook"
},
{
"slug": "why-now",
"label": "Why this question, now",
"kind": "prose"
},
{
"slug": "options",
"label": "The protocols",
"kind": "options"
},
{
"slug": "changes-mind",
"label": "What would change our mind",
"kind": "bullets"
},
{
"slug": "branches",
"label": "Where this leads",
"kind": "branches"
}
],
"images": [],
"claims": [
{
"id": "390065dca4fc",
"text": "Cloudflare reports its customers are already serving over one billion 402 responses per day, per the [pay-per-crawl announcement](https://blog.cloudflare.com/introducing-pay-per-crawl/).",
"status": "valid",
"lastChecked": "2026-05-06T19:49:39.314Z",
"note": "**Confirmed.** Cloudflare CSO Stephanie Cohen stated at CoinDesk's Consensus conference in May 2026 that Cloudflare's network processes a billion HTTP 402 'payment required' responses every day. The figure is reported by CoinDesk's coverage of the announcement (May 5, 2026). Note: the inline citation in this claim points at the original July 2025 pay-per-crawl announcement, which does not itself carry the daily-volume figure — that stat came from a later Cohen statement. Claim is accurate; the cited URL is suboptimal and could be updated to the CoinDesk piece."
},
{
"id": "0bff0c413d39",
"text": "Roughly 20% of Tollbit's network of nearly 7,000 publisher sites has earned revenue from the AI bot paywall, per [Tollbit / Adweek coverage](https://www.adweek.com/media/tollbits-publishers-monetize-ai-bots/).",
"status": "valid",
"lastChecked": "2026-05-06T19:49:39.314Z",
"note": "**Confirmed.** Adweek coverage states: 'Nearly 20% of TollBit's network of nearly 7,000 publisher sites has made money from its AI bot paywall, ranging from hundreds of dollars to tens of thousands a month.' Numbers and attribution match the claim verbatim."
},
{
"id": "bfdfe1fdb0e2",
"text": "ProRata reports more than 700 publication partners as of 2025, per its [Series B / Gist Answers announcement](https://www.businesswire.com/news/home/20250606852177/en/ProRata-AI-Signs-Partnerships-With-More-Than-500-Publications-Giving-Gist.ai-One-of-the-Largest-Licensed-Content-Libraries-in-Generative-AI-Search), including the FT, Axel Springer, The Atlantic, Fortune, Boston Globe and Vox Media.",
"status": "valid",
"lastChecked": "2026-05-06T19:49:39.314Z",
"note": "**Confirmed.** ProRata's June 2025 BusinessWire release was titled 'ProRata AI Signs Partnerships With More Than 500 Publications…'; subsequent reporting and ProRata's own materials updated the figure to 'more than 700 high-quality publications worldwide already participating'. Named partners (FT, Axel Springer, The Atlantic, Fortune, Boston Globe, Vox Media) are confirmed in WAN-IFRA, Press Gazette, and Digiday coverage. The 500→700+ growth between the announcement and the September 2025 Series B is consistent with the trajectory. Number stated in claim ('more than 700') matches public statements."
},
{
"id": "62f54bd3709b",
"text": "The FT–OpenAI deal is reported at \\$5–10M/year, per [Digiday's 2024 timeline](https://digiday.com/media/2024-in-review-a-timeline-of-the-major-deals-between-publishers-and-ai-companies/); the NYT–Amazon Alexa deal is reported at \\$25M, per [TechCrunch](https://techcrunch.com/2025/05/29/the-new-york-times-and-amazon-ink-ai-licensing-deal/).",
"status": "valid",
"lastChecked": "2026-05-06T19:49:39.314Z",
"note": "**Confirmed.** Both deal values are reported in their respective primary-source coverage: Digiday's 2024 timeline reports 'The Financial Times signed a deal with OpenAI worth $5 million to $10 million a year', and TechCrunch reports the NYT–Amazon deal at $25 million. Note: deal terms remain confidentially negotiated; figures originate from press reporting of leaked or partial-disclosure terms, not company filings. Use with the standard 'reported at' caveat."
},
{
"id": "34cee956dc78",
"text": "RSL reports 1,500+ media organisations, brands and tech companies endorsing the standard at launch, per the [RSL Collective announcement](https://rslstandard.org/press/rsl-standard).",
"status": "valid",
"lastChecked": "2026-05-06T19:49:39.314Z",
"note": "**Confirmed.** RSL Collective's launch press release (10 September 2025) states: '1500+ media organizations, brands, and technology companies worldwide now endorse RSL; support spans billions of web pages, representing most high-quality internet content used in AI foundation model training.' Number and framing match the claim."
},
{
"id": "97db563fdb65",
"text": "As of March 2026 x402 reports 119M+ transactions on Base and 35M on Solana, ~\\$600M annualised volume, per the [x402.org site](https://www.x402.org/) and [DWF Labs research](https://www.dwf-labs.com/research/inside-x402-how-a-forgotten-http-code-becomes-the-future-of-autonomous-payments).",
"status": "valid",
"lastChecked": "2026-05-06T19:49:39.314Z",
"note": "**Confirmed across multiple secondary sources.** As of March 2026, x402 reports 119M+ transactions on Base, 35M on Solana, and ~$600M annualised volume. Reported in Sherlock's x402 explainer, BlockEden's March 2026 piece, MEXC News, and Cryptonomist's January 2026 coverage. Note: the inline citation points at x402.org and DWF Labs; the live x402.org homepage does not currently surface the per-chain figures (they appear in Foundation/Coinbase blog updates and partner reporting), and DWF Labs is bot-blocked. Stat itself is accurate; the inline citation could be tightened to the BlockEden / Sherlock / MEXC pieces."
},
{
"id": "bed44d31517c",
"text": "Marketplace participation list and pricing are not yet public; only the launch partners are confirmed.",
"status": "valid",
"lastChecked": "2026-05-06T19:49:39.314Z",
"note": "**Confirmed by absence.** Microsoft's Publisher Content Marketplace was announced at an invite-only summit in Monaco (September 2025) with named launch partners (FT, Reuters, Axel Springer, Hearst, USA Today Network) per Press Gazette and Microsoft's own announcement. No public pricing schedule or full participation list has been disclosed; this is a meta-claim about the absence of disclosure, which is accurate as of May 2026."
},
{
"id": "e631778e8b19",
"text": "Per-publisher payout averages and conversion rates from Perplexity's revenue-share programme are not publicly disclosed.",
"status": "valid",
"lastChecked": "2026-05-06T19:49:39.314Z",
"note": "**Confirmed by absence.** Perplexity's publisher revenue-share programme has named partners (Time, Der Spiegel, Fortune, Entrepreneur, The Texas Tribune per Digiday's 2024 timeline) but no per-publisher payout averages or conversion rates have been disclosed publicly by Perplexity or partners. This is a meta-claim about the absence of disclosure, accurate as of May 2026."
}
],
"instructions": {
"forAgents": "Each section under `sections` is a structured component of this question. Treat `layer` + `question` as authoritative taxonomy. Cite by anchor: `/${layer}/${question}#${section}`."
}
}