AI Stole My Content and I Don’t Feel Good About It (Premium)

As a content creator, I see both sides of the AI debate. My experiences using Microsoft Copilot Pro have been nothing but positive, for example, and this product has saved me time and money. But generative AI’s need to suck up as much content as possible and then spit it back out without compensating the original creator is a troubling new development in the age-old battles over copyright and fair use. Indeed, it’s so troubling that it may require new laws and regulations.

This topic is curiously controversial with some, but this is my livelihood, so I see it as straightforward: Content creators should be paid for their work, and generative AI providers like OpenAI and Microsoft should pay them like Google Nes and other news aggregators do. There is no version of the term “fair use” that protects these firms: Arguments that generative AI is somehow “transformative” are immaterial given the massive and unprecedented scale of this theft of content and the financial reasons for it. Regardless, this isn’t about AI transforming copyrighted, protected content. It’s about AI consuming it without permission and then regurgitating it without compensation. That’s theft in the same way that selling a bootleg DVD on the streets of New York City is theft.

Tied to this, I have watched our revenues from ads collapse over the past year—they’re down to almost literally one-third of what they were a year ago—I assume it’s obvious to almost everyone that AI will only exacerbate this problem by not pushing their users through to the source material, which will impact traffic. Theoretical arguments about copyright, fair use, the rule of law, or whatever aren’t just uninteresting in this case, they’re insulting. This is a real problem, even if it does not impact you directly.

And now, like The New York Times, I have a bit of experience that makes the theoretical a reality. There’s an AI out there, one I’d not heard of previously, that sucked up at least one of my articles—I assume it scraped my entire site—and is spitting out my content in response to user prompts.

Or is it? Let’s take a look.

I received a very interesting email the other day from a person who noticed that one of my news articles was the first source of information for a prompt—Mozilla refocuses on AI—on Perplexity AI, a service with free and paid tiers that, among other things, offers what can only be described as a news feed. In fact, I suspect this is where the “prompt” I noted above came from. That is, it was likely automated by the system and not something a user got when prompting it.

Perplexity AI’s take on Mozilla refocusing on AI cites 7 sources across its 8 paragraph summary of this topic, and my article is the source of the first two paragraphs. This makes a head-to-head comparison of the two interesting. And, as it turns out, depressing.

The first sentence of the first paragraph of the AI summary reads:

Mozilla has announced a significant reorganization that includes layoffs and a reduction in the number of products it supports, with a new focus on integrating trustworthy AI into its Firefox web browser.

My version reads like so:

“Mozilla announced a reorganization that will result in layoffs, fewer products, and a new focus on trustworthy AI in its Firefox web browser.”

This is a simple rewording, the type of thing an unsophisticated, lazy blogger would do when translating a corporate blog post or press release into a shorter, less well-written blog post for their site. It’s arguably not even “AI,” in the sense that it’s just a shuffling of words.

It may also be worth pointing out that my article is based on a Bloomberg news story and a leaked version of the Mozilla memo that describes this shift. Two notes about that: My article is in no way a simple line-by-line rewording of someone else’s work, as I’m seeing with this AI regurgitation. And for AI to be truly effective, it should always use original sources of information as the primary source, in this case, the Mozilla memo. Stealing from me is two levels removed from the root of this story, and my writing includes biases and opinions that should be excluded from any AI summarization of a topic. (We see this problem every single day on Techmeme, as it rarely cites original sources when reporting on major news stories.)

The next line reads as:

This shift comes as Mozilla aims to prioritize resources in areas with the greatest chance of success, such as Firefox Mobile, and to establish a better model for the industry.

This one is interesting as it’s summarizing/rewording a direct Mozilla quote and not differentiating between my writing and that quote. That quote reads like so:

“We’re scaling back investment in some products in order to focus on areas that we feel have the greatest chance of success. We intend to re-prioritize resources against products like Firefox Mobile, where there’s a significant opportunity to grow and establish a better model for the industry.”

Again, just a simple rewording and not what I’d call AI. Though, AI is of course what did the rewording.

The next sentence, which starts the second paragraph, reads as follows:

Mozilla’s reorganization involves stepping back from certain projects, such as its Mastodon-based social media platform, consumer security and privacy products (Mozilla VPN, Relay, and Online Footprint Scrubber), and its Hubs 3D virtual worlds, due to a lack of interest and demand.

This is another rewording and summary, in this case of most of one of the paragraphs in my story:

“[Mozilla] is stepping back from its Mastodon-based mozilla.social social media platform and will instead ‘participate in the Mastodon ecosystem’ using a much smaller team. It is likewise walking away from its consumer security and privacy products—Mozilla VPN, Relay, and Online Footprint Scrubber (which was just announced as Mozilla Monitor Plus)—to focus instead on ‘products addressing customer needs in growing market segments.’ And it is winding down its work on its Hubs 3D virtual worlds due to a lack of interest and demand.”

That paragraph includes mostly original writing but also some direct quotes from the Mozilla memo. And so the AI summary above is a mishmash rewording of the two.

Instead, Mozilla is turning its attention to AI, driven by the rapid changes in the industry landscape and the potential for AI to reshape personal technology.

Is another mishmash of my writing and the Mozilla memo. In this case, I contributed:

“Mozilla, like the rest of the personal technology industry, sees AI as a potential savior.” It even moves the word “potential” into a sentence that was otherwise lifted from the memo:

“Generative AI began rapidly shifting the industry landscape … ”

The company sees an opportunity to bring trustworthy AI into Firefox, leveraging the Fakespot acquisition and subsequent product integration work.

That’s lifted almost verbatim from the memo, which reads:

“Mozilla seized an opportunity to bring trustworthy AI into Firefox, largely driven by the Fakespot acquisition and the product integration work that followed.”

Mozilla also plans to integrate its Pocket, Content, and AI/ML teams supporting content with the Firefox organization.

This is another example of AI summarizing/rewording, and is also taken from the memo:

“We will be bringing together Pocket, Content, and the AI/ML teams supporting content with the Firefox organization.”

Leaving aside the fact that a big chunk of the theft was taken from a memo I didn’t write—because the AI doesn’t know this and assumes everything there is fair game—I don’t see any added value in what the AI has generated. It’s just a simple rewording, as noted, the type of thing an unsophisticated and lazy blogger would do.

Put simply, for AI to be truly transformative in this case, it would need to add value in the form of context, additional information, or analysis. Rewording what I wrote is not transformative, it’s just theft. I’m also not being compensated for this theft, and, no, linking to me in a barely visible footnote-type fashion is no form of compensation. The goal of this product is not fair play, let alone fair use, it’s to keep readers on their site, steal content, and make money. Mission accomplished, it seems.

I look forward to the coming story about AI content theft in the Perplexity AI news feed.