No AI* Here – A Response to Mozilla's Next Chapter
Waterfox's founder sharply criticizes Mozilla's new AI-first strategy, arguing it compromises user trust and agency by integrating black-box LLMs into the browser. He contrasts this with Waterfox's commitment to a simple, user-controlled experience and auditable local ML. This resonates deeply with the Hacker News community's skepticism of corporate AI and a persistent desire for privacy-respecting, user-centric alternatives.
The Lowdown
Waterfox founder Alex Kontos has issued a pointed rebuke to Mozilla's newly announced AI-centric vision, challenging the direction of the Firefox browser and positioning Waterfox as a steadfast alternative. He expresses empathy for Mozilla's existential struggles but fundamentally disagrees with their approach to integrating Large Language Models (LLMs), arguing it undermines the very principles of user agency and trust Mozilla purports to uphold. The article meticulously differentiates between beneficial, transparent machine learning (like local translation) and the opaque, potentially untrustworthy nature of current LLMs within a browser context. It argues that a browser should be a 'user agent,' not an 'AI agent' that mediates and interprets the user's interaction with the web.
- Mozilla's Dilemma: The author acknowledges Mozilla's immense pressure due to declining market share and the rise of AI-driven browsers, necessitating revenue diversification beyond search.
- The 'AI' Asterisk: Kontos distinguishes between auditable, transparent machine learning tools (like Bergamot for local translation) and current black-box LLMs, asserting that the latter cannot be audited or trusted with user data.
- Browser's True Purpose: He posits that a browser is inherently a 'user agent,' acting on behalf of the user, and that introducing an opaque LLM layer fundamentally changes this relationship, creating a 'user agent user agent.'
- Trust and Transparency: Despite Mozilla's promises of optionality, Kontos questions how users can truly monitor or understand the behavior of a black-box AI deeply embedded in their browsing experience.
- Alienating the Base: The author suggests Mozilla's strategy risks alienating its core technical community and privacy advocates, who prioritize control and transparency, in a chase for mainstream users Chrome already dominates.
- Waterfox's Stance: Waterfox explicitly commits to not including LLMs, focusing instead on being a simple, performant, and user-controlled browser that respects web standards and user privacy.
- Importance of Governance: Kontos also highlights Waterfox's formal governance and legal entity, arguing this provides crucial accountability that many smaller, independent browser forks lack, especially concerning sensitive user data.
In conclusion, Kontos reaffirms Waterfox's unwavering commitment to empowering the user, serving as a bulwark against what he perceives as a misguided pursuit of AI integration that sacrifices user control and trust for perceived market relevance.
The Gossip
AI Ambiguity: Auditing Models and Data Egress
Many commenters challenged the article's distinction between 'good' (auditable, local ML like Bergamot) and 'bad' (black-box LLMs), arguing that LLMs can also be run locally and their black-box nature isn't unique. The core concern for many isn't the architecture itself, but whether the AI is local or cloud-based, implying data egress. They suggest the author might be conflating proprietary, cloud-hosted LLMs with the technology generally, and that a local LLM could address many privacy fears.
Mozilla's Misplaced Momentum: Alienating the Core
A recurring theme was widespread disappointment with Mozilla's historical decisions, with many users feeling the organization has consistently alienated its core technical user base. Commenters highlighted past unpopular features (like Pocket), perceived neglect of important projects (Servo, MDN), and a focus on chasing mainstream users with features that don't align with their community's values. The AI push is seen by many as another iteration of this trend, moving away from what makes Firefox unique and valuable to its loyal users.
User's Choice: The Push for 'No AI' Browsing
A significant portion of the discussion affirmed the desire for a browser free from integrated AI, or at least with highly controlled, optional, and local AI. Users expressed concerns about privacy, data usage, the potential for AI to make incorrect decisions (e.g., auto-filling forms), and the general 'bloat' that unwanted features bring. Many voiced that they use AI tools elsewhere but don't want them forced into their browser, echoing the article's call for user agency and a simpler, more focused browsing experience.