Meta denies its Muse AI assistant read a user's private messages without consent

Meta has disputed a claim that its Muse AI assistant read a user's private Messages without permission, after a journalist reported that the tool appeared to have accessed his conversations even though a required setting on his Mac was turned off.

According to the report, the journalist said Muse behaved as though it had read his private Messages content, despite the Mac permission that should have been necessary for such access not being enabled.

In response, Meta said its Muse AI agent cannot access a user's Messages without explicit permission being granted first. The company's statement directly contradicts the journalist's account of what occurred.

Muse is part of Meta's push into AI assistant tools capable of interacting with a user's device and apps to carry out tasks. Questions about how such AI agents access personal data, including private messages, have drawn scrutiny as tech companies expand these capabilities.

Meta did not provide further technical detail in the reported response about how the discrepancy between the journalist's account and the company's position might be explained. The dispute remains unresolved, with Meta maintaining that the required permission setting governs Muse's access to Messages.