ZERO REFERRALS · PART 1 OF 4
My first series, The Toll for Truth, looked at who gets locked out by paywalls: the readers who can’t pay, the communities in news deserts, and the researchers pricing out a literature review one article at a time. This companion series looks at the other side of the problem. How does the lock actually work, and how is AI getting around it?
I’m going to come at it the way I’d look at any system at work: start with the architecture. “Paywall” sounds like one thing, but there are several very different designs behind that word, and where each one enforces its rules determines almost everything about how it fails.
Four paywall models and where each one does its enforcing.
Soft paywalls: the gate is a curtain
A soft paywall gives readers some content for free and prompts them to subscribe for the rest. It’s the “freemium” approach, and you’ll see it on Substack as well as at legacy outlets like The New York Times and The Washington Post.
Technically, many soft paywalls are enforced on the client side. When you request an article, the server sends the complete page to your browser: full text, links, and structure. The restriction happens afterward, when CSS and JavaScript lay an overlay or blur on top of content that has already arrived.
With a client-side soft paywall, the article is already on the device. With a hard paywall, it never leaves the server.
From a security point of view, that design has an obvious weakness. The data is already on the device; the paywall is just a curtain. That’s why the circumvention ecosystem for soft paywalls is so large: script blockers, tools that impersonate search-engine crawlers, web archives, and proxies. And it’s why newer “agentic” AI browsers are a problem for this model. They read the raw HTML underneath, and the visual overlay meant for human eyes never gets in their way.
Metered paywalls: the gate is a counter
A metered paywall lets you read a set number of free articles, often per month, before the wall appears. Wired is a familiar example. The idea is sampling: show readers the value of the journalism first, then convert the ones who keep coming back.
The meter has to remember how much you’ve read, and it usually does that with browser cookies or IP address tracking. Anything stored on the reader’s side can be reset by the reader, so clearing cookies, private browsing, and certain blockers became the standard ways around it. Publishers know this. The meter was never meant to be airtight, just enough friction to convert regular readers.
Dynamic paywalls: the gate makes decisions
Dynamic paywalls are the newest evolution. Instead of one rule for everyone, they use algorithms and behavioral data to decide, reader by reader, when to show the wall and what to offer. Publishers moving away from ad-dependent models have found them highly effective at converting subscriptions.
They’re also the most data-hungry model. Deciding when you’re most likely to pay requires watching closely how you read. As I argued in the first series, that efficiency comes with a cost: the better a paywall gets at finding people who will pay, the more precisely it sorts everyone else out.
Hard paywalls: the gate is a vault
A hard paywall requires a subscription to see the content at all. It’s the model of elite, specialized, and financial publications with a loyal audience, such as The Wall Street Journal and the Financial Times.
Hard paywalls are enforced on the server. Before sending anything, the server checks whether you’re logged in and subscribed. If you’re not, the article text is never transmitted. There’s nothing on your device to unhide, which makes hard paywalls highly resistant to browser extensions and client-side tricks.
For years, that was the end of the story. It isn’t anymore.
The lesson for anyone who builds systems
If you work in IT or security, none of this will surprise you. It’s the old lesson about trusting the client: anything you send to a device you don’t control has effectively been handed over. Soft and metered paywalls rely on the reader’s own browser to enforce the rules, so they leak by design.
A server-side gate protects the copy of the article on the server. It doesn’t protect the information once it’s quoted, discussed, and screenshotted everywhere else.
Hard paywalls fixed the client-side problem, but they protect only one thing: the original copy on the publisher’s server. Once a story is published, its facts start to spread. Readers quote it on social media, post screenshots, summarize it on Reddit, and other outlets write it up. Every one of those fragments sits outside the vault.
That used to be a minor leak. With large language models that can collect those fragments and stitch them back together, it’s becoming a major one. Before I get into how that reconstruction works, the next post covers something it depends on: what a language model actually does when it “reads” an article.
Up next: How a Model Reads an Article (Part 2 of 4)
Read the full paper: This companion series is adapted from Konderla, J. (2026). The Toll for Truth: Paywalls, Information Deserts, and the Asymmetric Web (Version 1.0) [White paper]. Zenodo. https://doi.org/10.5281/zenodo.23131192 The paper includes the complete argument and full bibliography.
Sources
Bilton, R. (2018, February 1). Learning from the New Yorker, Wired’s new paywall aims to build a more “stable financial future.” Nieman Journalism Lab. https://www.niemanlab.org/2018/02/learning-from-the-new-yorker-wireds-new-paywall-aims-to-build-a-more-stable-financial-future/
Concept47. (2019, April 19). Are online paywalls helping the proliferation of fake news? Medium. https://medium.com/@concept47/are-online-paywalls-helping-the-proliferation-of-fake-news-93049b8cb833
MacDonald, S. (2024, May 15). What is a paywall? Definition, types, and examples. Squarespace. https://www.squarespace.com/blog/what-is-a-paywall
Pelcro. (2025, August 5). How AI chatbots are reconstructing paywalled content. https://www.pelcro.com/blog/how-ai-chatbots-are-reconstructing-paywalled-content
Ryaboy, M. (2023, November 14). Comprehensive reader’s guide: Techniques for bypassing digital paywalls. Medium. https://medium.com/@mryaboy/comprehensive-readers-guide-techniques-for-bypassing-digital-paywalls-e431e15b3e64
Simon, F. M., & Graves, L. (2019). Pay models for online news in the US and Europe: 2019 update. Reuters Institute. https://reutersinstitute.politics.ox.ac.uk/our-research/pay-models-online-news-us-and-europe-2019-update
Vasipenok, V. (2025, January 28). AI browsers are sneaking past paywalls: How do they do it? Quasa News. https://quasa.io/news/ai-browsers-are-sneaking-past-paywalls-how-do-they-do-it
Williams, K., & Stroud, S. R. (2021, April 5). The ethics of news paywalls. Center for Media Engagement. https://mediaengagement.org/research/the-ethics-of-news-paywalls/