Meta Muse Privacy: Surfshark Counts 31 Data Types
Published by AINave Editorial
Meta Muse’s privacy concern is less a single startling number than what the number measures. Surfshark’s comparison of Apple App Store disclosures lists 31 of 35 data types for Muse, just behind Meta AI at 33. That is a broad disclosure footprint, not a direct count of what the agent gathers from each person in daily use.
What the App Store comparison shows
Surfshark compared the listed data types for 13 widely used AI chatbots. In its tally, Google Gemini lists 24 types and ChatGPT 17, putting Muse ahead of both by this measure, but behind Meta AI. The comparison counts categories in disclosures; it does not establish how often information is collected, whether each category applies to every user, or what happens in a particular interaction.
| Chatbot | Data types listed, out of 35 |
|---|---|
| Meta AI | 33 |
| Meta Muse | 31 |
| Google Gemini | 24 |
| ChatGPT | 17 |
These are Surfshark’s counts of Apple App Store data disclosures, not independently measured collection volumes. The distinction matters: a disclosure category indicates the breadth of information described, but by itself does not show the amount collected or the circumstances in which it is collected.
Why an agent raises a different privacy question
The TechRadar account describes Muse as an agent intended to pursue tasks on a user’s behalf and says it may draw on information from calendars, emails, and bank accounts. Surfshark’s comparison also identifies sensitive categories, including ethnic data, sexual orientation, trade-union membership, political opinions, genetic data, and biometric data, among those listed by Muse, Meta AI, and Gemini.
The concern is the combination: personal information connected with interaction context could support detailed profiling. TechRadar also raises possible advertising or training uses, but the supplied reporting does not establish that these outcomes occur for every user or that the disclosure count proves them. Likewise, calendar or email references are not evidence of universal access to those sources.
That makes the ranking useful as a disclosure signal, not a verdict on what happens to an individual’s data. For an agent designed to act with personal context, the practical question is how access and data use work in the specific setup. The count alone cannot answer it.
