Meta Platforms is responding to claims that its Muse AI agent accessed private messages without authorization. The company asserts that the software requires multiple layers of user permission, but the incident highlights ongoing challenges in data privacy and user trust. Investors are monitoring how these privacy concerns impact the company's reputation and potential regulatory oversight for its AI products.
Meta Platforms is working to address concerns regarding its Muse AI assistant after reports suggested the tool accessed private user messages without explicit consent. In response, company executives have emphasized that the Muse software is strictly an opt-in feature, requiring specific manual configurations from the user before it can interact with personal data. According to Meta, the integration for Mac systems is designed with multiple layers of security to prevent unauthorized access.
Meta officials explained that users must manually enable a setting known as Full Disk Access and configure specific connectors within the Messages application. These steps create two distinct layers of user-authorized security. Furthermore, Meta representatives noted that the macOS interface requires users to navigate system settings to confirm these permissions, a process that triggers a restart of the application. This design, the company argues, makes it technically unlikely for the software to access private data without the user's full knowledge and intent.
The controversy surfaced after reports claimed the AI was able to access messages even when specific permissions were disabled. One suggestion was that the AI might have intercepted notification banners, reading text directly from the device screen rather than accessing the underlying database. Meta has dismissed this theory, suggesting that the AI’s explanation of its data source was likely a hallucination—a common AI error where the model generates incorrect or misleading information—rather than evidence of a security breach.
While Meta maintains its technical stance, the company continues to face a significant trust gap regarding its data handling practices. This incident follows a series of past legal and regulatory setbacks, including a verdict from a New Mexico jury that found the company had misled users regarding its data practices. Additionally, a separate incident has surfaced involving allegations from a content creator regarding the handling of sensitive information on Facebook Marketplace. Meta has acknowledged the possibility of a software error in the Marketplace case and confirmed that it is investigating the matter.
For investors, the long-term success of Meta’s consumer AI strategy is tied to its ability to maintain user trust and avoid further regulatory or legal scrutiny related to privacy. The ability to demonstrate transparent data controls and minimize software errors will be a key focus for market observers as the company continues to integrate AI across its platforms.