Mozilla is axing another project that felt a perfect fit for a company founded on the idea of making the web better for everyone.
Next week, on June 26 2025, the Fakespot Deep Fake Detector extension, which let users vet content to gauge if it was written by a human or generated by AI for a human, will stop working — news first reported by Soeren Hentzschel.
As an AI-powered tool, the Fakespot Deep Fake Detector wasn’t perfect at separating the wheat from the statistically-glued collage of plagiarised and paraphrased words, but it was at least powered by multiple open source models (ApolloDFT, Binocular, UAR, and ZipPy).
That “mean assessment” angle gave it a degree of utility that obfuscate pricey proprietary alternatives lack (and most online AI detectors are grifts to upsell access to their own generative model that can supposedly ‘bypass detectors’).
The end of the Firefox Fakespot Deep Fake Detector add-on will not be widely mourned since it had few only a modest ~3,300 active users. The web-based version, and the extensions for Google Chrome and Safari also go offline the same date.
When Mozilla announced the end of its AI-powered online review checker Fakespot1, the writing was on the wall for the ‘deepfake’ detector too.
Along with Fakespot, Mozilla has announced its closing the popular read-it-later service Pocket and nuking its own privacy-preserving AI assistant add-on Orbit from, well, orbit.
If it looks like a pattern and acts like a pattern… is it?
Pivoting to AI = pivoting away from us
Mozilla continues to make a big deal about how it’s pivoting to AI. This, it enthuses, will meet modern needs, secure the future of Firefox and ultimately push the web forward.
It says this while simultaneously shuttering the genuinely thoughtful, user-first AI products it has, like Fakespot, Fakespot Deep Fake Detector, and Orbit.
A mind more cynical mind might see Mozilla’s talk about ‘pivoting to AI’ as a half-finished statement that should end ‘pivoting to AI companies‘. Is the hope the big players will pay to see their tech integrated in Firefox (and ensure Mozilla CEO pay continues to buck industry trends </sass>)?
I’m not usually one for “hot takes”—online discourse is a sweltering mess of them—but Mozilla’s recent actions (about ‘saving money and better investing resources’, it will say) do suggest its new-found obsession with AI is revenue-centric, not user-centric.
Kills its own AI efforts to integrate others’
Orbit, the privacy-preserving AI assistant add-on ran used an open-source LLM that processed data on Mozilla’s own servers rather than a third party’s, was fast, and didn’t require signing up or signing away data.
If you’d asked 2017 me to sketch out how the Mozilla I grew up respecting would handle the era of AI chatbots, Orbit would be it.
Yet it was axed in favour of adding third-party AI company chatbot to the Firefox sidebar. Sure, we as users get to pick our fave but is that “user choice” when “choice” is a selection of sandwiches filled with… Ahem. I’ll go hungry.
Admittedly, Firefox’s AI chatbot sidebar stuff in Firefox is not (yet) enabled by default, but its “integration” feels user-hostile.
I toggled on ChatGPT when needing to illustrate something in a recent article. Now, every time I select text on a web page, an annoying on-screen chip and chat box appear, forcing it on me.
Which is why I cycle back to this point of who this “pivot to AI” is for.
Making basic text selection trigger an AI prompt modal is annoying and clearly there to game engagement. Cue: “lots of people use {LLM} via Firefox’s sidebar, so what you say {LLM maker}: wanna pay us to keep it there?”.
If we as users want “summaries” to avoid reading; “hyper-realistic” imagery to share on social media to reap indifference; school work to offload the effort of learning; soliciting life ‘advice’ from statistical models programmed to please… It’s everywhere anyway.
Mozilla isn’t making any of this stuff “easier” to access in Firefox since it’s all a tab away anyway. Should Mozilla be spending its supposedly-constrained resources building in third-party services, or should those services be doing via add-ons?
Worse, these chatbots operate like adult phone line workers in the ’90s. They say what you want to hear and reinforce your own beliefs and ideas to make you feel big, feel heard, feel good. Dopamine-inducing pleasure that hooks you into using it more.
Less web browsing, less critical thinking, more pleasure seeking, more token rate exceeding ‘sign up for a pro plan now to keep feeling good’.
Killing off Deepfake—a tool which ostensibly frames AI-generated content as lesser than human content—is a sign Mozilla views AI-generated content (or “commodi-tent”, since no-one uses AI to pump out 400 articles a day for noble reasons)—as being worthy.
But why would a company built to champion the open web be willing to kill of its own AI projects that aid the open web, in favour of adding other people’s products that are seeing the walls close in on the web as we’ve known it?
Firefox is barely relevant. It’s not a player in the AI space, its marketshare is in free falls. An atrophic web, hollowed out by AI-generated slop, hurts Firefox. A wounded Firefox spells trouble for Mozilla.
Mozilla may think its survival now depends on serving the needs of big AI companies rather than its own users or indeed its own interests.
—AI companies who foretell, with profit-driven purpose in their eyes, the oncoming end of the human web and the ambrosian utopia that awaits from having all human knowledge, interaction, and creativity mediated via their products.
In killing off a tool it designed to help humans tell human from machine, has Mozilla made clear whose side it is on?
Only a cynic would think stuff like that…
- Talking of, a company called hCaptcha reportedly approached Mozilla with an offer to buy Fakespot, pledging to keep it running as a public service, sans ad monetisation or data collection. Mozilla’s answer? It ignored them – principles, eh? ↩︎