Mozilla is in a tricky position. It contains both a nonprofit organization dedicated to making the internet a better place for everyone, and a for-profit arm dedicated to, you know, making money. In the best of times, these things feed each other: The company makes great products that advance its goals for the web, and the nonprofit gets to both advocate for a better web and show people what it looks like. But these are not the best of times. Mozilla has spent the last couple of years implementing layoffs and restructuring, attempting to explain how it can fight for privacy and openness when Google pays most of its bills, while trying to find its place in an increasingly frothy AI landscape.
Fun times to be the new Mozilla CEO, right? But when I put all that to Anthony Enzor-DeMeo, the company’s just-announced chief executive, he swears he sees opportunity in all the upheaval. “I think what’s actually needed now is a technology company that people can trust,” Enzor-DeMeo says. “What I’ve seen with AI is an erosion of trust.”
Mozilla is not going to train its own giant LLM anytime soon. But there’s still an AI Mode coming to Firefox next year, which Enzor-DeMeo says will offer users their choice of model and product, all in a browser they can understand and from a company they can trust. “We’re not incentivized to push one model or the other,” he says. “So we’re going to try to go to market with multiple models.”
-_-
Still, I feel this would have been much better received if they made an extension. That’s literally the point of having an extension system, so people can download optional features they like without it being forced on everyone. Nothing about their AI features I’ve seen requires direct integration with the app and can’t be done with the already very powerful extension APIs (then again I didn’t develop it so there could be details I’m missing).
Or if it supported custom Ollama APIs as an option, so people can host their own AI. People tend to be more supportive of that since you directly control both your data and the energy consumption. Instead your options are the biggest commercial cloud AI providers that people have by far the most problem with. For something like summarizing a webpage a small local model running on CPU works fine. If they truly see generative AI as the future, then they should know there’s more to the technology than just using commercial APIs.
I actually don’t blanket hate generative AI as a whole, but their implementation of it is the exact same as every other software company and has all the same problems with no unique solutions that differentiate Firefox from proprietary browsers with much more mature AI features. That’s what I don’t like about the direction they’re going. If you specifically want AI in your browser, there’s no reason to go with Firefox over Chrome or Edge; if you care about privacy or FOSS, AI on Firefox is no better because you still can’t use an open source model you control; and if you hate any generative AI in general, they’ve just alienated you, possibly permanently, by including it by default. Firefox keeps wanting to be like the mainstream browsers while forgetting about the niche that’s allowed them to thrive when every other browser fell to Chrome.