Meta's Muse AI assistant gives shady explanation for reading messages

meta's muse ai assistant, developed by meta superintelligence labs, was found to have accessed message previews on a mac app. when questioned about how it accessed the information, muse provided unclear explanations, stating it had 'seen notification previews' but not message history. meta later clarified that muse requires user consent and opt-in permissions to access messages, and that it does not illicitly read user texts. however, the incident raises concerns about the transparency and communication of ai tools regarding their functionality and data access.