Meta's New Terms Make Publisher Content Free by Contract — and Publishers and Advertisers Just United to Object

Meta's updated Commercial Terms take effect on October 30, 2026, and they convert continued use of a Facebook, Instagram or WhatsApp business account into consent for automated crawling of the domains that business is associated with. The grant is explicit: Meta may use scrapers, bots and crawlers, create test credentials, and access, index, analyse, collect, retrieve and crawl those domains, pages and URLs — including the content available through them — in order to \"provide, develop, improve, research, and maintain advertising and business products.\" Turns a business account into a licence to crawl the open web

Two protections are missing. There is no carve-out and no contractual promise to honour robots.txt; Meta says owners can still signal preferences that way, but Benjamin Sobel of the University of Wisconsin points out that a voluntary convention is not an enforceable preference, and a law-firm alert on the change calls robots.txt \"a request, not a locked door.\" And the word \"train,\" present in the version published on October 15, was deleted over the weekend, with permitted uses narrowed to advertising and business products. Matt Topic of Loevy & Loevy says \"develop\" is broad enough to include training; Cornell's James Grimmelmann says what remains is \"broad enough to drive a truck through.\" Deletes the word 'train' without narrowing what the data can be used for

Six days turned a contract edit into a coordinated European campaign. On Monday, October 19, the European Publishers Council wrote to EU competition commissioner Teresa Ribera and digital commissioner Henna Virkkunen, asking Brussels to suspend the terms before Friday and to test them against EU copyright and competition law. On Tuesday, the German Association of Digital Publishers and the Free Press Association — together representing more than 9,000 media brands — asked the Commission to block the change, arguing it makes the \"free surrender of the entire digital publishing product\" the price of access. On Thursday, the Movement for an Open Web, whose membership counts both digital advertisers and publishers, filed complaints with the Commission and the CMA and asked for interim measures. Turns a contract change into a three-city campaign in six days

The alignment has a mechanical cause, and that is why this is not simply another publishers-versus-platforms row. The clause does not reach only publishers' journalism; it reaches the website any advertiser, merchant or agency keeps for its own business. One grant sweeps up both sides of the same counterparty: the content a publisher sells and the product data, catalogue and audience signals an advertiser holds. Both groups are being asked to trade a priced market for a compulsory contract, and both would rather pay a visible fee than sign an unlimited one. Makes one clause bind the publisher's journalism and the advertiser's product data at once

The timing is the part investors should get right. The CMA cannot order an interim suspension through its Digital Markets Act powers because Meta has not been designated as a firm with strategic market status. Chief executive Sarah Cardell told members of Parliament on Tuesday, October 20, that the watchdog had written to Meta, met it on Wednesday and was engaged in \"urgent discussions,\" and she named Meta as a candidate for designation while conceding the options are limited without it. General competition enforcement remains open, the Commission has confirmed it is \"engaged\" but has opened no case, and the Movement for an Open Web's request for interim measures requires it to show \"serious and irreparable damage.\" Puts the October 30 deadline beyond the reach of either watchdog

The strategic shift underneath is what makes this worth more than a privacy grievance. By moving the practice into a contract, Meta moves the argument away from copyright law — where the fight over training on public web content has been running for years — and into the fairness of a gatekeeper's terms for business users, which is the provision the German publishers actually invoked. Copyright questions are slow and jurisdictionally messy; a gatekeeper's contract terms are the kind of thing competition authorities are built to police. Moves the dispute onto the ground where competition law bites fastest

The clause attacks a market that exists only because AI developers had no cheaper way to buy content. OpenAI has signed agreements with roughly 35 publishers, Perplexity's publisher programme covers about 20 outlets, and trackers counted 94 publicly announced AI content licensing deals by July 2026. The disclosed prices are real money: News Corp's five-year agreement with OpenAI is estimated at about $250 million, Thomson Reuters has booked $33 million of AI licensing revenue, People Inc. receives at least $16 million a year from OpenAI, and Wiley reported $49 million of AI revenue in fiscal 2026. Puts a visible price on content the platform's own contract now treats as free

None of this needs the clause to be struck down to matter. If one platform can write into a contract that content from any linked site is available for its advertising and business products, every other buyer has to justify paying $50 million a year for what a competitor treats as free. That is an anchor effect, and it lands first on the people currently being paid. Caps what any AI buyer can pay for content without a court ruling

Meta reported the third quarter on October 21, 2026, and the numbers frame the decision: revenue of $64.6 billion, up 26%; advertising revenue of $63.1 billion, up 26%; capital expenditure of $34.9 billion in the quarter and $135–145 billion guided for the year; and free cash flow of negative $2.1 billion. Funds an AI build that makes free data a strategic necessity

None of this is about saving the licence bill. Even if Meta paid every major publisher the $50 million-a-year rate it pays News Corp, ten such deals would cost about $500 million a year — roughly a third of one percent of $135–145 billion of annual capital expenditure. What the clause buys is not money but legal architecture: it pre-empts the question of whether AI companies must pay for content at all, and it does so in the one document a business cannot realistically refuse to sign. Buys legal certainty about consent rather than a cheaper data bill

The mechanism runs through advertising and, increasingly, through the assistant. Permitted uses in the clause point at \"advertising and business products,\" and the quarter shows the prize: ad impressions rose 10% while the average price per ad rose 14%, with nearly 10 million advertisers already using at least one of the platform's generative creative tools. On the earnings call the company said more than 10 million people had signed up to its Muse assistant in the US and Canada within six weeks, that more than a quarter of users use it to shop at an average purchase above $200, and that it plans to charge businesses fees on the transactions it facilitates. Shopping assistants need product, price and availability data — precisely what a merchant's website holds. Pushes the ad and shopping advantage toward data the platform takes for nothing

Snap is the cleanest read-across because it has the least room to answer. Revenue grew 19% to $1.60 billion in the second quarter, but advertising revenue grew only 9% and the operating loss was $171 million, with a $324 million operating loss over the trailing twelve months. It licenses no content and holds no contractual right to crawl, so when a rival converts product-level data into a 14% higher price per ad, the only defence is audience scale — and it does not have its rival's. Leaves Snap with the least room to answer a rival that owns the data layer

Alphabet is the counterweight and, in practice, the standard setter: $119.8 billion of quarterly revenue and $40.8 billion of operating income mean it can buy any licence it wants, and its search business monetises queries built on publisher content every day. TikTok, owned by ByteDance, faces the same incentive and is already a designated gatekeeper in the EU, so its version of the clause is less likely to be treated as a novel contract question. In the United States the response is loud but formal: trade groups have called the change \"probably illegal,\" and no US agency has announced a case. Turns a European contract dispute into a template that Washington will have to answer

Over one to three years the question is not whether the clause survives intact. It is whether \"the web is free for AI\" becomes the default commercial assumption. If it does, the publisher content market compresses toward the zero price the platform has published, Thomson Reuters's $33 million of licensing revenue becomes a rounding error inside a $7.8 billion revenue base, and every licensor negotiates against an anchor it did not create. If regulators or courts hold that commercial terms cannot manufacture consent to give content away, the opposite happens: content keeps a price and the developers that currently pay for it can keep paying. Reprices the raw material of AI whether or not the clause ever reaches a court

The evidence supports three different levels of confidence. Facts: the clause exists, takes effect on October 30, carries no opt-out and no contractual robots.txt commitment, and has been met with coordinated complaints in Brussels, Berlin and London by publishers and advertisers together. Inference: a visible zero price for web content will compress what AI developers pay licensors even if the clause is never challenged, because it moves the reference price. Speculation: the platform may ultimately pay for content it now takes, if a court or the Commission forces the point — the cost would be trivial against $135–145 billion of annual capital expenditure, and the precedent would not be. Separates what the documents prove from what the market should price