If you run a newsroom or a magazine, or you write for a living, you’ve probably felt the new version of piracy:
Your work appears on someone else’s website, or as LLM answers.
The reader gets what they need.
Your site never gets the visit.
No bueno.
The thesis I keep coming back to is something that Steve Jobs and Daniel Ek proposed about music, when building iTunes and Spotify respectively:
The only way to win is to make permission easier than piracy.
Courts matter. Deals matter. Standards matter.
But “permission beats piracy” only when it’s simpler, faster, and less risky for the buyer.
This became abundantly clear in the Bartz v Anthropic copyright lawsuit in the latter half of 2025, when it emerged in the case that Anthropic executives knew what they were doing when they decided to skirt the law and use pirated material.
They called the effort to license the works they wanted to use as “business slog”, as it was a tedious and difficult task to individually license the various body of works from the multitude of fragmented libraries and rights-holders.
It’s not excusable, but it gives an insight as to why they made that decision: it was too painful, time-consuming, and costly to take the legal route, and there was no simple alternative.
Again, not excusable, but you can see the logic.
Why “pirate first” stops working (even for companies that can afford lawyers)
AI companies have three problems that don’t go away:
Enterprise customers hate risk and uncertainty. If you’re selling into banks, pharma, governments, or big brands, “we might be sued” is a deal-killer, bringing unnecessary risk.
Discovery is expensive and ugly. Lawsuits turn training data and system logs into a liability and a distraction.
Quality has a ceiling. If the inputs are messy, the output won’t feel trustworthy, and trust is what businesses pay for.
So yes, I expect more licensing. Not because AI firms suddenly develop a conscience. But more so because licensing becomes the cheaper, simpler path compared to years of uncertainty.
But that’s not the end of the story.
There’s an uncomfortable pill for publishers to swallow, which is not too palatable:
Even if AI labs pay, it does not automatically restore your old economics.
Clicks used to bring ads, subscriptions, and user-habits.
But a cheque or a deal doesn’t bring back the clicks.
They’re gone; most likely for good.
So the goal can’t be “win a deal and relax.”
The goal is: build a permission product.
The “better than piracy” lesson from music (and what it means for publishers)
Steve Jobs’ and Daniel Ek’s core point (paraphrased) was simple: you won’t defeat piracy by trying to shame people with morality or trite aphorisms; it doesn’t work. People value simplicity, and will use the simplest option available to them.
Instead, you defeat piracy by offering something better that still pays the rights-holders.
That logic is now walking into publishing.
But “better than piracy” for written content doesn’t mean another paywall or robots.txt.
It means making permission feel like ordering from a menu, not negotiating a treaty.
What “permission” looks like as a product (simple enough to copy this week)
If you publish reporting, analysis, or a valuable archive, your permission product has four parts:
1) A clear terms page (one page).
Plain language. No legal fog.
What’s allowed, what isn’t, and how to license.
Simple.
2) Pricing tiers (a menu, not a mystery).
Keep it boring and obvious. Example tiers:
Reference + attribution (quoting short excerpts with a source line)
Answer-use (summaries in responses, with a source line)
Training-use (ingestion for model training / fine-tuning)
Archive / bulk access (licensed feeds, back-catalog packages)
Different value. Different price. Different rules.
3) A fast permissions process (speed is the advantage).
A single inbox or form. A promise like:
“We reply within 48 hours.”
If you don’t reply, the buyer learns a bad habit: “scrape and move on.”
Speed and simplicity go hand in hand.
4) Machine-readable terms (so the bots can read your rules).
This is the only technical bit, and it can stay simple:
a label that says, clearly, what’s permitted and what isn’t, in a way automated systems can understand.
That’s it.
No hype. No complicated stack. No lawsuits.
Just a permission product that’s easier to use than scraping.
It’s worth repeating, so I will:
The only way to win is to make permission easier than scraping.
Where provenance and blockchain fit in (only if they reduce friction)
Most people, publishers and journalists included, hear “blockchain” and immediately think of Bitcoin, NFTs, and the hype train. Fair enough.
But ignore the label and focus on the function:
proving who created what (and when),
showing what rights apply,
recording usage,
paying out in a way that can be audited by anyone.
You don’t need a grand theory. You need receipts. Blockchain is how you get them.
If a buyer can’t tell you what they used, when, and under what terms, you don’t have a real market; you have a recipe for disaster.
What publishers and freelancers can (and should) do this quarter
If you’re a publisher:
Turn your archive into inventory. List your durable assets (evergreen explainers, investigations, specialist beats, datasets, images). Decide what you’re willing to license. I suggest to start small, and expand as you go.
Publish the one-page terms + menu. If it takes more than 3 minutes to understand, it’s not a product. It’s a problem.
Separate answer-use from training-use. Don’t let them blur together. Different value, different risk, different terms.
If you’re a freelance journalist:
Stop treating reuse as a favour. Create your own simple licensing terms for your work (even a lightweight version).
Ask for clarity in contracts. Who can license your work to AI partners? Do you share upside? Do you keep any rights? What happens for re-use?
Make yourself easy to cite. A consistent author page, a portfolio, clean attribution language.
And for everyone else who writes for a living:
Permission isn’t a moral argument. It’s a sales funnel.
Your job is to make it easy to do the right thing, and expensive to do the lazy thing.
Because (one last time, everyone together):
The only way to win is to make permission easier than piracy.
If you want to roll this out across your publication, or for your freelance work, book a short 15-minute call with me. We’ll define what you will license, what you won’t, and who owns approvals, so you don’t get shafted, or stuck in internal limbo.
Happy New Year everyone!
Around the Bloc is back after a brief pause. We’ve got a brilliant agenda lined up for you for 2026!
If you enjoyed this, you’ll definitely enjoy next week’s article:
“The Attribution Gap: why AI can use your work, why you can’t prove it (yet) + the fix”.
Stay tuned, and make sure you’re subscribed to get each issue sent straight to your inbox.