.png)
2 hours ago
1
Machine Learning: General
People keep asking for ways to disable AI. Just yesterday in this reddit thread: https://www.reddit.com/r/firefox/comments/1o1mnrf/comment/nihqshp/
People end up posting lists of preferences to flip like this:
user_pref("browser.ml.enable", false); // machine learning features in Firefox user_pref("browser.ml.chat.enabled", false); // AI Chatbot (https://docs.openwebui.com/tutorials/integrations/firefox-sidebar/#additional-about-settings) user_pref("browser.ml.chat.menu", false); // "Ask a chatbot" in tab context menu user_pref("browser.ml.chat.shortcuts", false); // "Enable custom shortcuts for the AI chatbot sidebar" user_pref("browser.ml.chat.shortcuts.custom", false); user_pref("extensions.ml.enabled", false); // might only be relevant for app developers user_pref("browser.ml.linkPreview.enabled", false); user_pref("browser.tabs.groups.smart.enabled", false); // "Use AI to suggest tabs and a name for tab groups" in settings user_pref("browser.tabs.groups.smart.userEnabled", false)We should make sure to give them an option to do this. Otherwise people will accidentally break things in weird ways and it will look like a bug on our part.
We should probably focus on remote services because for some definition of AI we use AI even for detecting forms on pages but that has nothing to do with LLMs and it does not send any user data anywhere.
Type: defect → enhancement
We have an Enterprise Policy that tries to do that without disabling our offline machine translations that is more focused on GenAI https://mozilla.github.io/policy-templates/#generativeai
Someone want to create a quick table of the prefs that we have and what they do and then we clean it up such that there's a global "Gen AI" pref that Firefox (and enterprise policies) can use before the individual features are checked?
![Dark Patterns: Are Your Games Playing You? [video]](https://www.youtube.com/img/desktop/supported_browsers/firefox.png)
