Privacy

VxNewton for iPhone

VxNewton generates responses on your iPhone using Google's open-weight local AI model Gemma. This page explains privacy as it relates to VxNewton.

What is the model, and how is it running on my phone?

VxNewton includes its AI model Google Gemma in the bundled app download. The AI model is simply a file with lots of numbers and code to generate text; this is effectively what all LLM AI models are.

Saved chats

VxNewton does not send your conversations to a remote AI/LLM data center. VxNewton computes all of the responses locally on your iPhone using Google's Local AI model Gemma. When a prompt requires a web search for an optimal response, VxNewton may use a privacy-focused DuckDuckGo web search tool to give you up-to-date accurate information, while maximizing privacy. Search queries generated by VxNewton are sent to DuckDuckGo, not your prompts or specific chat messages.

You can delete a chat from the sidebar. Your saved chats and settings may be included in iCloud or computer backups, depending on your settings. Deleting a chat in VxNewton does not remove copies from previous backups.

Web search

Web Search is on by default in VxNewton because this is necessary for VxNewton to provide a reliable default experience to all users. When VxNewton decides it needs to do a web search, it will generate its own search query and look up the query on DuckDuckGo, and this does not send your messages, prompts or chats to DuckDuckGo.

You can turn off Web Search in Settings.

Contact

For privacy questions, email support@vxnewton.com.