THE RICHARD J MURPHY YOUTUBE CHANNEL
DEBATE AMMUNITION
THE AI CON TRICK
Funding the Future | May 2026
TODAY'S TOPIC
AI Is an Economic Con-Trick. The video relating to this content is here.
THE CORE ARGUMENT
Artificial intelligence companies have built their businesses by scraping vast quantities of other people's content without permission and without payment and they then charge fees to access it. That is not innovation; it is rent extraction in its most classical form. Unless governments act now to regulate AI charges, tax AI profits where the economic impact arises, and enforce the property rights of content creators, a tiny number of corporations will hold permanent, unchallengeable power to charge the rest of the world for access to what might be their own knowledge.
KEY STATISTICS
| Statistic | Figure | Source |
|---|---|---|
| Number of blog posts written by Richard Murphy over approximately 20 years, all potentially scraped by AI companies without permission or payment | More than 25,000 | Video script (approximate figure stated by Richard Murphy) |
| Number of videos produced by Richard Murphy, whose transcripts are freely available online and similarly exposed to scraping | More than 1,000 | Video script (approximate figure stated by Richard Murphy) |
| Amount of compensation received by Richard Murphy from any AI company for use of his content as training data | Zero | Video script |
| Permission sought by any AI company from Richard Murphy before using his content for commercial training purposes | None | Video script |
THE ARGUMENT STRUCTURE
Step 1 — The business model is extraction, not creation
Large language models do one thing well: they take vast quantities of other people's data without paying for it, process it, and sell access to the result. The output may appear coherent, but the underlying model is one of appropriation, not genuine origination.
Step 2 — This is a breach of copyright and licence conditions
Richard publishes under a Creative Commons licence that permits non-commercial use but requires permission for any commercial application. AI companies appear to have used his work with the clear intention of making a profit, without seeking that permission. He argues this constitutes a breach both of copyright and of the explicit licence terms under which his material is made available.
Step 3 — The correct name for this in political economy is rent extraction
A rent is an access fee charged for an asset by someone who claims the right to own it, regardless of whether they created it. AI companies scraped the web voluntarily, assembled databases from other people's work, and now charge subscriptions to access them. The analogy is precise: the content creators stand in the same position as the land in a tenancy arrangement, the asset that generates all the value receives none of the return. Those who enclosed the property collect everything.
Step 4 — The political stakes are without precedent
Tony Blair's vision of AI as our economic future means a massive redistribution of wealth away from ordinary people and towards the owners of AI platforms. The inequality that unchecked AI rent extraction could produce may dwarf everything that has come before it, including tax injustice, financialisation, and wage suppression. This is why democratic accountability, fair taxation of AI profits at source, and enforceable creator rights are urgent, not aspirational.
THEIR ARGUMENT → YOUR REBUTTAL
| They Say | Your Response |
|---|---|
| AI companies add enormous value through their processing and engineering. They are not simply reselling other people's work; they have built something genuinely new. | Nobody disputes that processing has a value. The question is whether the underlying data they process was theirs to use. It was not, always. A landlord who builds a house on land they have enclosed without consent is still guilty of enclosure, whatever they build on top of it. The value of the processing does not extinguish the claim of the person whose raw material was taken without permission. |
| Web scraping is legal, and the courts have consistently found it falls within fair use or equivalent exceptions. There is no copyright violation here. | The legality of scraping in general is one question; whether using scraped content for commercial profit without licence permission is a separate and much live one. Richard publishes under a Creative Commons licence that explicitly prohibits commercial use without permission. No AI company sought that permission. Legal cases are already before the courts for precisely this reason, and the outcome is not settled. Law has always been written to protect established property interests; that does not make the arrangement just. |
| If you do not want your content used, you should not publish it on the open web. Publishing online is an implicit consent to its use. | Publishing under a Creative Commons licence is the opposite of blanket consent. It is a carefully specified set of conditions, including the explicit restriction on commercial exploitation. The argument that publication implies consent to any use someone can find technically possible is the same argument enclosers made about common land: that unused land was, by implication, available for private appropriation. We know how that story ended. |
| AI will create enormous economic growth and new jobs. We should not strangle it with regulation before it has had the chance to deliver those benefits. | The promise of future general prosperity is the standard argument made for every new form of enclosure and rent extraction throughout economic history. The distributional question is not whether AI generates aggregate output; it is who captures that output. A system in which a handful of corporations hold the rights to charge the rest of the world for access to knowledge is not a growth story. It is a monopoly story, and it demands urgent regulatory action, not patient deference. |
THE ONE-LINER
“AI companies scraped the web without permission, built databases from other people's work, and now charge us all to access it: that is not innovation, it is enclosure, and we should say so.”
FURTHER READING
| Post Title | Date | Relevance |
|---|---|---|
| Debate ammunition: The Pope's AI encyclical | May 2026 | Companion briefing on Pope Leo XIV's encyclical Magnifica Humanitas, which frames AI as a question of power and democratic accountability, directly complementing this video's political economy argument. |
| Blair is claiming TINA. I am sure there is TIARA | May 2026 | Richard's critique of Tony Blair's AI essay, which this video references directly as representing the pro-AI redistribution-upwards agenda. |
| The problem with AI is it just does not care | Jan 2026 | Argues that AI developed primarily as a private asset will deepen inequality, extract rents, and undermine democratic control, echoing this video's central thesis. |
| Will AI create a new digital divide and drag the economy down? | Jan 2026 | Examines the distributional and macroeconomic consequences of AI's dominance, including the subscription rent model and its effect on those without regular income. |
| When AI, unemployment, and inflation collide | Jan 2026 | Identifies systemic risk from overinflated AI company valuations and the macroeconomic consequences of job displacement driven by AI rent-seeking behaviour. |
| AI won't save us | Oct 2025 | Directly challenges the mainstream narrative that AI will deliver broad economic salvation, supporting the argument that AI's benefits are concentrated in very few hands. |
| The problem with AI | Aug 2024 | Earlier formulation of Richard's core concern that AI represents a form of extraction rather than genuine value creation, with regulatory failure at its heart. |
| We can use AI in the economy but only if we are willing to tax capital more heavily | May 2023 | Early statement of the tax justice principle applied to AI: that any economic gains from AI must be redistributed through heavier taxation of capital, or inequality will compound. |
Thanks for reading this post.
You can share this post on social media of your choice by clicking these icons:
There are links to this blog's glossary in the above post that explain technical terms used in it. Follow them for more explanations.
You can subscribe to this blog's daily email here.
And if you would like to support this blog you can, here:
Buy me a coffee!
This took me back to the early noughties and one of my hobbyhorses, the NoDPI campaign against Phorm Inc, ISPs, & digital companies, involving unlawful interception of communications including 3rd party web content that visited, + scraping of content for profiling & delivering targeted advertising – for commercial gain. The ISPs claimed “deemed consent” from the creators of the web content being scraped because it was “freely available”. Similar processes were (and still are) used for webmail, both for targeted profiling and for detecting malware in email. As a campaign member & customer of BTBroadband and owner of 2 scraped websites, I got involved in lengthy legal arguments with a top legal honcho at TalkTalk including seeking counsel’s opinion (all in my archives).
One of the less successful arguments we made was about copyright protection for website owners who were third parties in the ISP/ISP customer contract. Even when we added specific copyright clauses to our websites forbidding this scraping, their legal advice was that they could scrape anyway. So we diverted certain ISP’s customers to explanatory pages, as part of the campaign – but obviously that wasn’t any way to generate website traffic.
The copyright issue never got to court, we couldn’t afford to take on BT, Virgin Media and TalkTalk on DPI/website copyright, let alone G**gle and Yahoo (over webmail scraping without consent of third party recipients of our email messages).
Good luck with the copyright arguments. I think you are right, but unless you have a sympathetic millionaire to finance a civil case, it won’t run. As for expecting prosecutions by corrupt law enforcement system in the UK – CoL police/DPP not interested – same with local forces, even escalating as far as possible.
Thanks
These cases are running