Our minds on media.

Are AI companies violating copyright when using online material for training. Aren’t search engines? Aren’t you?

Just to be clear, I’m not calling you stupid. That’s a reference to the old Bill Clinton campaign war room sign, “It’s the Economy, Stupid.” A rallying cry for his campaign for president in the 1992. It was an effort to keep campaign staffers on message and to not be distracted by more superfluous issues. A.I. is not exactly a superfluous issue, but the idea that using material from the web to train A.I. or allowing A.I. to search and summarize for users, and not compensating anyone for it; well, that’s a problem.

Current copyright laws do not protect facts. They don’t protect the labor utilized to create new facts. If I report here that genetic engineering may be creating new species of superweeds, I am not violating copyright.¹ I’m not even really “reporting it” as far as I’m concerned. I’m giving you a link or two so you can go see for yourself, and then I can proceed to editorialize on the matter. Regardless, a new problem has arisen that is the same as the old problem. We really have no current way to value content on the web. We really never have. There are two significant side effects to that: the twin plagues of advertising and zombie web sites.

The prevalence of advertising on the web says it all. The medium is free but people want to (need to) collect some revenue for the content they produce. Since web consumers largely got used to not paying for content on the web, paywalls and business models like them didn’t succeed very well. Advertising has not just succeeded, it is the driving force on the web. Google, frequently described as a search engine company, is really an advertising agency. In 2024, Alphabet, Inc. (Google’s parent company) reported revenues of $350 billion and of that, advertising revenue from it’s various services (Google Search, Youtube, and Google Network) amounted to about 75% of its revenue or $262.5 billion. I’m picking on Google, but most social media companies look roughly the same. ²

Most people don’t see this as a bad thing. It is bad. As demonstrated by the superweeds footnote here, the information you get for free rather depends on who benefits from generating it. That’s not usually you. Incentivizing content creators to gather “eyeballs” and “foot traffic” has lead to a general attention economy that has been bad for information health and even our mental health. Clickbait is a term we all know because of hyperbolic headlines that are false on their face.

Then there are the zombie web sites (my term). These are sites that are built with borrowed or outright stolen content to sell advertising space or manipulate search results. Buzzfeed went out of business in 2023 and yet… it’s still there, generating content somehow without much of a staff and collecting ad revenue. At any rate, Buzzfeed aside, SEO experts speculate that anywhere from 10-30% of indexed web sites have the characteristics of low-quality, auto-generated content. Noise, really. (It’s far worse with email—at about 50-60% spam—as we all know.)

What A.I. is doing is not something that search engines aren’t already doing. To pick on Google again, nearly 1/3 of searches on google’s search platform end without a click, meaning that the person doing the search likely found the fact that they wanted and didn’t move on to the source. (So much for critical thinking.) Regardless, news media bargaining laws have been enacted in a few places (Australia, Canada) and there is an old proposal in the US (Journalism Competition and Preservation Act) which would enable media organizations to negotiate prices for search engines to display their content. You and I can’t. Just them. That’s not very much in the spirit of the web’s design.

We need an actual marketplace for content. I’ve advocated for a long time that the web needs a money protocol. To use it, you pay a microfee³ in the neighborhood of $0.0001 or smaller to visit a web site. The protocol would be optional. We have http (HyperText Transfer Protcol), and a secure version in https. We need httpm (or httpc for currency). The wild west of the free web would still exist, there would just be sites that would charge for entrance—not via a cumbersome subscription and paywall—but just a small transaction fee. Bots would have to pay. A.I. companies would have to pay. It would incentivize content creators to be produce better information, not just sexier information.

All the parts for such an endeavor are there. The inventor of the web, Tim Burners Lee is working on a project called Solid that would allow web users to control their information. This effort could incorporate microfees. Web3 is an effort to build a decentralized version of the web on blockchain technology. It could do the same. Some browsers (like Brave and Opera) now have cryptocurrency wallets built in to them, although none of this has to necessarily involve cryptocurrency. The point is, you deposit some money in your browser, arrive at a ‘httpm’ destination and agree to pay and enter, or don’t.

If media and tech companies (and us little creators) don’t work on a project like this, we are going to get more law suits regarding copyright and more dysfunctional regulations for corporations that disregard the independent creator all together. The web is a public space, yes, but like many examples of the tragedy of the commons like overfishing and traffic congestion, creating a toll-version of the web may be the best way to protect it and benefit creators.