Waterfox positions itself as an AI-free Firefox alternative

The Firefox fork Waterfox is positioning itself as a conscious counter-movement to Mozilla's AI strategy. As Waterfox developer Alex Kontos announced in a blog post, the browser will not contain Large Language Models – "period," as he unequivocally put it. The announcement was a direct response to Mozilla's plans to equip Firefox with AI features.

In mid-December, under its new CEO Anthony Enzor-DeMeo, Mozilla announced its intention to further develop Firefox into a "modern AI browser." The strategy includes features such as an AI Window where users can choose between different language models, as well as features like "Shake to Summarize" for iOS. Following massive criticism from the community, Mozilla developer Jake Archibald promised an AI switch that is intended to completely remove all corresponding functions.

In his blog post, Kontos distinguishes between specialized, transparent machine learning tools like Mozilla's translation tool Bergamot and generic LLMs. While purpose-built ML tools are auditable and traceable, he criticizes LLMs as intransparent black boxes. He sees potential access to browser data as particularly problematic: LLMs with access to tabs, history, and user interactions could manipulate content, alter displays, or filter out sensitive data.

Technical differences to Firefox

Waterfox is based on Firefox ESR and thus follows a longer update cycle than Mozilla's monthly releases. The browser, originally launched in 2011 as a 64-bit rebuild of Firefox, disables telemetry data collection by default, which Firefox claims to use for stability improvements, and offers features such as a vertical tab bar. A Classic variant maintains compatibility for older extensions that no longer work in modern Firefox versions.

The technical implementation of an AI kill switch remains unclear meanwhile. Mozilla has not yet disclosed whether the announced switch merely hides UI elements or also blocks network calls and telemetry to AI endpoints. Whether Firefox will rely on local or cloud-based models by default was also left open by Mozilla. For privacy-critical applications, a complete deactivation of all AI processes and network connections would be necessary.

GDPR relevance for AI integration

If Firefox transmits data to cloud LLM providers outside the EU, this is subject to the GDPR. Mozilla would have to ensure a legal basis, such as the necessary consent, and secure data transfers through standard contractual clauses. German data protection authorities could initiate investigations for large-scale AI functions involving personal data. Data exports to the USA, in particular, pose additional requirements.

For companies and public institutions that must rely on privacy-compliant browsers, Mozilla's final technical implementation will be crucial. Waterfox is likely to benefit from the debate and could become an alternative for users who fundamentally reject AI functions, alongside other forks like LibreWolf.

(mki)