HomeBlogThe Samhub SignalDigital Robber Barons: when Might makes Right

Digital Robber Barons: when Might makes Right

Digital Robber Barons: when Might makes Right

Meta's new terms for publishers are not the disease. They are the symptom. Martin Bergqvist on the digital robber barons of the AI economy, and what publishers can do now.
Samhub Signal #12: Digital Robber Barons, when might makes right

Table of Contents

Subscribe to our newsletter
Summarize this article with AI:

The AI economy is built on high quality journalism and content. The people building it have decided they don't need to ask.

30 October 2026 was supposed to be the day.

On that date, every publisher with a business account on Facebook, Instagram, WhatsApp or Threads would have "agreed" to let Meta send its crawlers through their websites. Meta would collect, cache and analyse what it found, and use it to "develop" its products. Nobody asked you to sign anything. If you kept running your campaigns, that counted as a yes.

The date has moved. In two weeks, publishers on both sides of the Atlantic pushed back hard enough that Meta blinked. A court in Oslo went further and stopped it.

I want to use this edition to say something I think the industry keeps tiptoeing around. Meta's terms are not an accident, and they are not a one-off. They are a symptom. The disease is a business model that treats independent journalism as free raw material for the AI economy. It is now being backed by a political climate where the strongest player simply takes what it wants.

In the 1880s, people had a name for this. They called them robber barons.

An unsigned contract on a desk, symbolising terms publishers never agreed to

What Meta actually wrote

Meta's updated Commercial Terms apply to anyone using its platforms for business: running ads, managing a Page, using Ads Manager or Business Suite. That is every publisher I know.

In the first version, you authorised Meta to use "web scrapers, bots, robots, spiders, crawlers" on the domains "associated with" your business. Meta could then use any content found there to "provide, develop, improve, research, train" its products.

When the word "train" caused an outcry, Meta deleted it. The content can now be used to "develop" Meta's "advertising and business products". US media lawyer Matt Topic put it simply: "develop" is broad enough to include "train".

Meta says you can still block it with robots.txt. But publishers have to let Meta's ad crawler in, or their own campaigns stop working. That makes it a choice between your journalism and your ad revenue. As Danielle Coffey of the News/Media Alliance said: "Meta will still say: take it or leave it."

Here is how the last two weeks played out:

• 15 October: the terms become public. The News/Media Alliance calls them "outrageous and probably illegal". The UK's News Media Association calls them a "shameless landgrab".
• 16 October: Meta deletes the word "train".
• 22 October: WAN-IFRA and the News/Media Alliance send letters. The European Publishers Council, German, Spanish and pan-European publisher associations file complaints with the European Commission.
• 23 October: Meta delays the rollout to 20 November.
• 27 October: ten Norwegian media companies, including NRK, Schibsted, Amedia, Polaris, TV 2 and NTB, take Meta to the Oslo District Court through their association MBL.
• 28 October: the court issues a temporary injunction. Meta may not implement the terms until their legality is decided. The same day, Meta says it will consult industry groups and may delay beyond 20 November.

In Sweden, TU and the Swedish Union of Journalists have taken the issue to the Swedish Competition Authority.

This is a win. But it is a pause, not a retreat. Apart from one word, Meta has not withdrawn anything.

A reporter taking notes at a local council meeting

The AI economy runs on your newsroom

Why would one of the world's richest companies fight this hard for the right to crawl regional news sites? Because that is where the value is.

Large language models are only as good as what they learn from. Most of the web is noise: spam, duplicates, unverified claims, content written to game an algorithm. What the models need is the opposite. Text that is accurate, edited, sourced and up to date.

That is a description of journalism.

The numbers back it up. A 2023 study by the News/Media Alliance found that the curated datasets behind popular LLMs overweight publisher content by a factor of 5 to almost 100, compared with the open web. In Google's C4 training set, half of the ten most-represented sites were news outlets.

Think about what sits behind every one of those articles. A reporter at a council meeting on a Tuesday evening. An editor who sends a story back because a source does not hold. A fact-check, then another one. A legal review. Years of trust built with a local community.

That is expensive. It is also exactly what makes an AI answer worth trusting. The AI economy has found a way to capture that value without paying for what produces it.

And it is not only about training. Answer engines now summarise the article instead of sending the reader to it. A Pew Research study found that Google users clicked a traditional result 15% of the time without an AI summary. With one, that fell to 8%. Our journalism trains the machine, then the machine stands between us and our readers.

A giant container ship passing a small wooden boat at sea

When might makes right

For most of my career, the digital economy had rules, even if they were imperfect. Copyright meant something. Competition law applied to everyone. If a platform crossed a line in Europe, the Commission could fine it, and that was the end of the discussion.

That world is coming apart.

In July 2025, when the White House launched its AI Action Plan, President Trump said it plainly. AI companies cannot be expected to pay for "every single article, book or whatever" they learn from. "You just can't do it. China's not doing it."

That is not a legal argument. It is a statement of power: we are big enough, and the race is important enough, that your rights are in the way.

The same stance shows up whenever Europe tries to enforce its own rules. When the Commission fined Apple, Meta, Google and X, Washington answered with threats. In December 2025, the US Trade Representative said it would use "every tool at its disposal". It named European companies such as Spotify, SAP and Mistral AI as possible targets. The same month, the US imposed a visa ban on former EU commissioner Thierry Breton.

I am not writing this to take sides in American politics. I am writing it because the message to European media is clear. The old order, where the rules applied to the strong as well as the weak, is no longer something we can take for granted.

In that world, Meta's terms make perfect sense. Write the rules yourself. Make acceptance automatic. Dare anyone to say no. Might makes Right.

A railway baron overlooking tracks built across the landscape

The digital robber barons

The original robber barons owned the railroads and the oil refineries. They did not produce the goods that moved through their networks. They controlled the routes, set the prices, and squeezed everyone who had no other way to market. It took a new kind of law, the Sherman Antitrust Act of 1890, and two decades of fighting before Standard Oil was broken up in 1911.

Today's version controls a different kind of infrastructure: the feeds, the search results, the ad auctions and now the AI assistants. At Samhub we call them Global Media. Meta, Google, OpenAI and the rest of the AI platforms all share the same business model:

Comparison of two business models: Publishers and Media versus Global Media

For twenty years, Global Media used our journalism to drive engagement on their platforms and took the advertising budgets that used to fund it. Now the AI players have adopted the same model: take the content, create value from it, and pay nothing, or pay a few selected publishers on terms they set alone.

That last point matters. Meta has signed AI licensing deals with a handful of large US publishers. Paying some while writing terms that let you take from everyone else is not a market. It is divide and rule.

The price is not only commercial. When newsrooms close, information bubbles grow. Trust in edited, accountable journalism erodes. And the AI answers we increasingly rely on are built on an ever thinner layer of verified facts.

People working together to pull a boat ashore

Five things to do before the next deadline

The Norwegian publishers have shown that the strongest player does not always win when the rest of us act together and use the rules we still have. Here is what I would do now.

1. Map your exposure. List every Meta business account you run and every domain "associated with" it: main sites, local editions, campaign pages, newsletter landing pages. That is the surface Meta's terms claim access to.
2. Reserve your rights, in writing and in code. EU copyright law lets you opt out of commercial text and data mining, but the reservation has to be machine-readable. Update your website terms to forbid scraping and AI training without a licence. Then back it up with a robots.txt that separates ad crawlers from AI crawlers, user agent by user agent.
3. Watch who is actually visiting. Monitor your bot traffic. You cannot negotiate over value you cannot measure.
4. Act together. No single publisher can out-muscle Meta. Associations can, and so can competition authorities. Support MBL in Norway, TU and Konkurrensverket in Sweden, and the European Publishers Council in Brussels.
5. Reduce your dependency. Every lever Global Media pulls works because publishers rely on their platforms to reach readers and advertisers. The way out is to own the relationship: logged-in readers, first-party data, and direct deals with advertisers who know what your audience is worth.

The fifth point is the hardest, and the one that lasts. Terms can be delayed and courts can grant injunctions. But the only thing no platform can rewrite is a direct, consented relationship between you and your audience.

The common thread

The robber barons of the 1880s were not stopped by goodwill. They were stopped when the people they squeezed organised, and when lawmakers decided that being big was not the same as being right.

We are at that point again. The AI economy cannot exist without what independent media produces: verified, edited, accountable information. That gives us more leverage than we think, but only if we stop handing it over for free.

A more profitable media business means more investment in high-quality journalism. And that matters for democracy, now more than ever.

So here is my question to you. If Meta's terms had come into force today, how many of your domains would it have been allowed to crawl? And what would it have cost you to say no?

If you do not know the answer, write to me at martin@samhub.io. I would like to hear how your house is handling this.

Samhub was founded by six Swedish media houses that got tired of watching Global Media take what was theirs. We were built by media, not just for media. There's a difference.

Win back what's yours.

Until the next edition.

Subscribe to Samhub Signal

Get the latest news for publishers & media straight to your inbox. One issue every other week.

Subscribe to Samhub Signal

More from the blog

  • Platform
  • Solutions
  • Resources
  • Pricing