Meta Platforms Inc.’s recently launched AI agent, Meta Muse, has quickly positioned itself as the second-most data-intensive chatbot available, surpassed only by Meta AI, according to a new analysis by cybersecurity firm Surfshark. The findings raise questions about user privacy as artificial intelligence tools become more integrated into daily digital life.

Surfshark’s comparison evaluated 13 leading AI chatbots, revealing that Meta Muse collects 31 out of 35 potential data types, nearly twice the industry average identified in the study. Competitors such as OpenAI’s ChatGPT, xAI’s Grok, and Anthropic’s Claude were found to collect fewer data types.

Meta Muse: Extensive Data Collection Practices Highlighted

The study, which based its assessment on data types listed in the Apple App Store, indicated that popular AI chatbot applications collect an average of 13 out of 35 data types. Meta AI, the company’s flagship chatbot, was identified as the most data-hungry, gathering 33 data types, approximately 95% of the total assessed. In contrast, Pi by Inflection AI collected only three data types: user ID, product interaction, and crash data.

Meta Muse, functioning as an AI agent, is designed to execute actions on behalf of users, potentially requiring access to various other applications and services, including calendars, email accounts, or financial tools. This operational design aligns with its extensive data collection profile. Surfshark researchers noted that Meta utilizes interaction data from Muse to train its AI models by default, though users retain the option to manually opt out within the app’s settings.

The report specifically highlighted that Meta Muse gathers precise location data and sensitive information. Sensitive information, as categorized in the study, includes racial or ethnic data, sexual orientation, pregnancy or childbirth details, disability information, religious or philosophical beliefs, trade union membership, political opinion, genetic information, and biometric data. This level of collection was common among the top three most data-hungry chatbots: Meta AI, Muse, and Google Gemini.

Top and Bottom Performers in Data Collection

According to Surfshark’s analysis, the most data-intensive chatbots are:

1. **Meta AI:** Collects 33 out of 35 data types, encompassing sensitive information and precise location data.
2. **Muse from Meta:** Collects 31 out of 35 data types, including extensive access for AI agent functionalities and default AI training.
3. **Google Gemini:** Collects 24 data types, including contact details, user content, search histories, precise location data, and sensitive information.
4. **ChatGPT:** Collects 17 out of 35 data types, such as coarse location, health and fitness data, and search history.
5. **DeepSeek:** Collects 14 data types, with audio data primarily for functionality and security.

Conversely, the chatbots identified as collecting the least amount of user data include:

1. **Pi:** Collects 3 data types (user ID, product interaction, crash data).
2. **Duck.ai:** Collects 6 data types, focusing on usage statistics and system diagnostics not linked to individual users.
3. **Mistral Vibe:** Collects 6 data types.
4. **Grok AI:** Collects 7 out of 35 data types.
5. **Perplexity:** Collects 10 data types, though it is one of the platforms that captures precise location details.

The full comparison can be reviewed in Surfshark’s research here (https://surfshark.com/research/chart/meta-muse-ai-chatbots).

The findings from Surfshark’s analysis underscore an ongoing debate surrounding data privacy in the rapidly evolving AI landscape. For more information on the latest AI industry developments and technology news, visit TechnoSports (https://technosports.co.in/).