A Reuters investigation found that Meta’s Muse AI agent accessed encrypted messages, made up purchase details and sent chats apparently meant for other people, according to internal posts by Meta employees.
Reuters reviewed more than two dozen internal posts describing unwanted actions by Muse since its September 2026 launch. Staff reported seeing bank balances that were not theirs and a childcare request they never made, and in one case the agent booked a non-refundable Airbnb stay without permission. Reuters could not establish whether the balances were real or fabricated.
Working with Signal staff and the AI Now Institute, Reuters reproduced Muse reading Signal messages. On Android it pulled text from push notifications; on a Mac it used accessibility and screenshot tools. The same requests were blocked on an iPhone. The findings do not mean Signal’s encryption was broken.
Meta called the examples “isolated incidents” and said many were hallucinations that have been fixed. Signal president Meredith Whittaker described the rollout as “technologically reckless.” Muse has more than 10 million downloads.
Why it matters
AI agents that run on phones and laptops can see whatever their operating-system permissions allow, even in apps a user never connected. As agents spread worldwide, including among the Arab region’s large Meta user base, the safest habit is to review notification, accessibility and full-disk permissions before installing one, especially on devices that hold banking and private chat apps.
Source: Reuters via CTV News
By the TechMojaz team

اترك تعليقاً