AI-generated slop is already clogging the arteries of the web
PLUS: Will Spotify's audiobook streaming be good for authors?
Welcome! I'm Simon Owens and this is my media industry newsletter. You can subscribe by clicking on this handy little button:
Let’s jump into it…
AI-generated slop is already clogging the arteries of the web
Ask any content creator or media executive what worries them about the rise of generative AI, and almost all of them will cite its potential to replicate their content and siphon away their audience.
These anxieties have sparked countless debates about the parameters of copyright protection, and they’ve led to dozens of lawsuits against companies that vacuum up web content to train their LLMs. The most high profile of these lawsuits came from The New York Times in December. It accused OpenAI of “unlawful copying and use of The Times’s uniquely valuable work” so that it could use it “without payment to create products that substitute for The Times and steal audiences away from it.”
But while these concerns are legitimate, much of the harm against publishers is still hypothetical, in that most haven’t seen a measurable hit to their business that can be attributable to the existence of AI chatbots. All you have to do is look at the NYT’s most recent quarterly earnings report to understand that it’s still growing unhindered. We haven’t seen the same kind of revenue fallout that followed the rise of, say, Napster, which very quickly began eating into music labels’ profit margins.
Meanwhile, I think there’s a much more pressing problem whereby AI is already starting to degrade the entire user experience of the web, rendering once-reliable content resources as completely useless. This is a problem that’s currently causing web users harm while simultaneously destroying the utility of the world’s largest tech platforms.
To understand the risk here, it’s helpful to look back at the pre-Google internet — an era when there were nearly a half dozen search engines vying for users’ attention. At the time, search engines like Yahoo, Lycos, AskJeeves, and AltaVista didn’t have very effective methods for indexing and sorting the web. Their ranking algorithms were fairly rudimentary; for instance, they placed significant weight on the number of times a particular keyword was used on a web page.
Because of this, search results were easily gamed and therefore useless. I’m speaking from experience here; I remember trying to conduct research on them for school projects and being frustrated when they returned absolute drivel. They were so bad that a company called Dogpile gained significant audience share for its sole function of aggregating results from multiple search engines — the idea being that no search engine by itself was up to the task of surfacing what you wanted.
We all know what happened next: Larry Page and Sergey Brin invented Pagerank, a much more elegant algorithm for ranking web content, and as a result Google captured nearly all the search engine market share. Within a year or two, web users began to trust search engines to serve up (mostly) relevant results.
Flash forward to today, and that trust is quickly eroding — not just for Google, but also all sorts of tech platforms that service billions of users.
The first victim was Amazon’s book store, likely because it provides the easiest way for AI-aided scammers to generate revenue from their work. As Wired reported in September, its search results have been flooded with AI-generated knock-offs, mostly in reference categories like travel and how-to:
There’s a whole strain of hustle culture YouTube videos encouraging people to get rich quick by churning out AI-generated works for Kindle, with titles like “How to Make UNDETECTABLE AI Content for Amazon KDP (Guaranteed Method).” Unfortunately for regular people trying to buy books online, this new moneymaking scheme makes it harder to find high-quality books written by humans.
I’ve seen multiple comments from social media users who said they accidentally ordered these books from Amazon, and in each case they immediately were able to tell that the book was written by AI. Despite these bad customer experiences, Amazon told Wired it has no current plans to scan for and identify AI-generated books.
Then there’s Google News, which, as 404 Media reports, is indexing sloppy, AI-generated rewrites of already existing news articles:
One example was a news site called Worldtimetodays.com, which is littered with full page and other ads. On Wednesday it published an article about Star Wars fandom. The article was very similar to one published a day earlier on the website Distractify, with even the same author photo. One major difference, though, was that Worldtimetodays.com wrote “Let’s be honest, war of stars fans,” rather than Star Wars fans. Another article is a clear rip-off of a piece from Heavy.com, with Worldtimetodays.com not even bothering to replace the Heavy.com watermarked artwork. Gary Graves, the listed author on Worldtimetodays.com, has published more than 40 articles in a 24 hour period.
It’s worth pausing for a moment to note the significance of this. Google News has strict parameters for what it allows into its index and requires new publishers to fill out an application to join. The parameters are so tight that I haven’t even bothered to submit my own newsletter, despite the potential upside of being indexed.
And what is that upside? Well, in addition to showing up in Google News search, articles are often displayed in a special “top stories” module at the very top of search results. They’re also featured within the mobile home page when over a billion Android and iOS users open their Chrome apps. It’s clear that websites indexed by Google News are given much more authority and prominence throughout all of its apps. I’ve worked at media companies that regularly received traffic windfalls of tens of thousands of visits in a few hours just because a story ranked well in Google News. And despite the importance of Google News in our information ecosystem, a Google spokesperson confirmed to 404 Media that it has no current plans to identify and label AI-generated content in search results.
Finally we have the wide proliferation of scammy ads that are displayed across millions of websites, social media platforms, and video portals via programmatic ad tech. This allows AI-generated, misleading slop to appear even on legitimate news sites:
Late last year, Scarlett Johansson sued a company that used her synthetic likeness to market AI content-generator apps. Around the same time, a disparate range of celebrities were compelled to warn fans about online clones. Gayle King encountered herself reading an ad for a weight-loss product she’d “never heard of” or used. “There’s a video out there promoting some dental plan with an AI version of me,” wrote Tom Hanks on Instagram. Mega-popular YouTuber MrBeast cautioned his followers not to fall for convincing but fake iPhone-giveaway ads. On YouTube and even X, Elon Musk replicas are still promising untold riches in exchange for a little bit of private information and — no big deal, don’t worry about it, it’s just a formality — some banking details.
Again, it’s not hyperbolic to conclude that this content is already generating real-world harm, potentially against hundreds of thousands of web users. Despite this huge problem, I’ve yet to come across an ad tech platform that’s developing ways to detect and eliminate this fakery. In fact, many ad tech vendors merely brag about how brands can leverage AI in their campaigns.
Right now, we’re still in the first phase of this AI-generated internet dystopia. For the most part, billions of users are still trusting these platforms to deliver high quality information. But remember the examples of the pre-Google search engines; as consumers continue to get burned by AI-generated slop, it’ll trigger a growing distrust of the platforms’ ability to aggregate and serve up trustworthy information.
The cynics reading this will respond by saying they gave up trusting the big tech platforms long ago, but while those platforms certainly have a checkered history of content moderation, I think those mistakes will pale in comparison to what’s coming if they don’t do a better job of detecting and filtering AI-generated content. This isn’t a hypothetical harm that’s still a few years off in the distance; this slop is already clogging the arteries of the web, and it’s quickly going to get worse.
What do you think?
How Gabe Fleisher built Wake Up To Politics, a daily newsletter with over 50,000 subscribers
There are plenty of popular politics newsletters out there, but none with the kind of unique origin story of Wake Up to Politics. It was started by Gabe Fleisher when he was only 9 years old. While the early editions were sent out by a Gmail account and only read by his mom, Gabe kept at it, waking up early every day before school to write the newsletter. Flash forward about a decade, and he’s now a senior at Georgetown and Wake Up to Politics has close to 50,000 subscribers.
In my interview with Gabe, we talked about what kept him motivated all these years, how he monetizes the newsletter, and what he plans to do with it once he graduates.
Watch our interview in the video embedded below:
If video embeds don’t work in your inbox, go here.
If you want to listen to an audio version of this interview, subscribe to The Business of Content wherever you get your podcasts: [Apple] [Spotify] [Amazon Music]
I’m looking for more media entrepreneurs to feature on my newsletter and podcast
One of the things I really pride myself on is that I don’t just focus this newsletter on covering the handful of mainstream media companies that every other industry outlet features. Instead, I go the extra mile to find and interview media entrepreneurs who have been quietly killing it behind the scenes. In most cases, the operators I feature have completely bootstrapped their outlets.
In that vein, I’m looking for even more entrepreneurs to feature. Specifically, I’m looking for people succeeding in these areas:
Niche news sites
Video channels like YouTube, TikTok, and Instagram Reels
Podcasts
Newsletters
Affiliate/ecommerce
Interested in speaking to me? You can find my contact info over here. (please don’t simply hit reply to this newsletter because that’ll go to a different email address. )
Will Spotify's audiobook streaming be good for authors?
It’s been nearly five years since Spotify announced it would diversify its audio offerings beyond music streaming, and while it spent most of that time building its podcast capabilities, it made no secret that it eventually wanted to get into audiobooks.
Then in 2022 it made its first move into the industry by acquiring an audiobooks distributor called Findaway. Later that year, it launched the ability to purchase audiobooks through Spotify. And then finally in late 2023 it rolled out audiobook streaming as part of its paid subscription.
There’s one group that’s watched these developments closely: audiobook authors. They’re understandably nervous about how Spotify’s bundled offering will affect their own income, and many are deeply skeptical of the company’s intentions.
So will Spotify’s audiobook streaming be good for authors? That’s a question I put to Jane Friedman, the writer behind the publishing industry newsletter The Hot Sheet. She walked me through the current landscape of digital audiobook sales and explained how Spotify’s revenue sharing arrangement works.
Watch our discussion in the video embedded below:
If video embeds don’t work in your inbox, go here.
If you want to listen to an audio version of this interview, subscribe to The Business of Content wherever you get your podcasts: [Apple] [Spotify] [Amazon Music]
Quick hits
I have a hard time believing Hollywood can just consolidate its way out of its current predicament. More acquisitions will just mean these companies are bogged down with more debt. Most of them have great offerings, they just need to focus on doing the hard work of improving their customer acquisition process, generating more revenue per user, and decreasing churn. [Variety]
Not long ago, there was some hopeful optimism that a handful of tech billionaires could bring some fresh thinking to the challenge of properly monetizing news content, but they're just as flummoxed as everyone else on how to revitalize legacy media outlets. [NYT]
INTERESTING: LinkedIn is launching a way for you to email gate your content within LinkedIn. Basically you can publish an article on LinkedIn and then insert a "unlock article" that forces them to input their email address before they can access the full thing. [Social Media Today]
"Over a full fiscal year, The Guardian received roughly $30 million in revenue from reader donations in the U.S. and Canada. About a third of all digital reader revenue across The Guardian now comes from the United States." [Nieman Lab]
The headline's a little hyperbolic, but I do agree that allowing writers to sell one-off information products like ebooks and courses could be a big opportunity for Substack. [Storyletter]
Want to pick my brain on your content strategy?
Are you contemplating a new content strategy and want someone to give you feedback? I’ve had more and more of my readers reach out and request consulting calls so they can pick my brain on a variety of issues including platform optimization, content strategy, and monetization.
You can now book a call with me directly through Substack. Use the link below to grab a time on my calendar:
Book a meeting with Simon Owens
Want a daily dose of media industry news?
I only send this newsletter out twice a week, but I curate industry news on a daily basis. Follow me on one of these social platforms if you want your daily fix: