Friends stop each other from feeding data to AI agents
agents meta
| Source: The Verge | Original article
A surge in consumer gadgets that indiscriminately collect data is prompting concerns as AI assistants like Meta’s Muse request personal information.
Meta’s new AI personal assistant, Muse, is drawing scrutiny after researchers revealed that the service can amass data about people who never signed up for the platform. By tapping into a user’s contact list, messages and social interactions, Muse builds “personal files” on friends, family, former dates and anyone else the user has communicated with – even if those individuals never gave consent.
The finding, reported by The Verge and highlighted by Cybernews, notes that up to 3.9 billion Meta accounts could be feeding the assistant’s memory with third‑party information. Meta counters that each user’s Muse data lives on a dedicated cloud computer, and a spokesperson told Wired that an AI agent needs contextual knowledge of a user’s contacts to fulfil requests effectively.
Why it matters is twofold. First, the practice blurs the line between personal data and the data of by‑standers, raising fresh privacy concerns for a platform that already handles massive amounts of personal information. Second, it spotlights a broader trend in consumer tech where devices and services harvest ambient data without explicit permission from all parties involved, potentially inviting regulatory scrutiny under emerging data‑protection laws.
What to watch next includes Meta’s response to the criticism – whether it will introduce opt‑out mechanisms for non‑users, tighten data‑minimisation safeguards, or face formal investigations from privacy regulators. Industry observers will also be looking for how other AI assistants handle third‑party data, as the debate could shape standards for consent and transparency across the rapidly expanding AI‑assistant market.
Sources
Back to AIPULSEN