AIToday
Large Language ModelsAI Safety & AlignmentThe Verge AIPublished: Sep 11, 2026, 04:00 JST2 min read

Meta's Muse works but crawls personal data, tester finds

Meta's Muse works but crawls personal data, tester finds

3 Key Points

  1. What happened

    Meta launched Muse, an AI assistant that handles emails, shopping, and trip planning. Reviewer Emma Roth found it deleted thousands of emails and bought workout tops, but it also gleaned detailed personal data.

  2. Why it matters

    Muse pulled her interests from Instagram's API data, not the UI, and inferred her location from an Amazon shipping address. This shows the assistant accesses more personal information than users can see, raising privacy concerns.

  3. What to watch

    Meta says Muse only shares data needed for third-party apps and doesn't share with advertisers. The real test is whether users trust it with credit cards and emails given Meta's privacy track record.

WHO IT HITSPrivacy-conscious users considering agentic AI assistants, especially those who link social media and shopping accounts, may hesitate to adopt Muse. The article suggests that suspicion could be a hurdle for Meta to overcome.

Ask the AI about this article →

Summaries like this, in your inbox every morning.

Context & Analysis

Meta's launch of Muse marks its first real foray into AI-powered productivity tools, a shift from a company that previously prioritized entertainment. The assistant carries out actions using a cloud-based virtual computer, and in testing, it successfully deleted thousands of promotional emails and purchased workout tops with consideration for items already in the cart. However, the experience was overshadowed by the amount of personal information Muse autonomously gathered.

Muse pulled specific interests from Instagram's API data, which surfaces more than the app's user interface, and it admitted to using an Amazon shipping address to infer the reviewer's location. Meta says it designed Muse to only exchange data with third-party apps that's needed for them to work, and doesn't share information with advertisers. Yet the reviewer's unease reflects a broader tension: as AI agents take on more personal tasks, the boundary between helpful automation and invasive data collection becomes harder to define. The outcome likely hinges on whether Meta can convince users that Muse's access is limited and transparent, especially given its privacy track record.

FAQ
What tasks can Meta's Muse assistant do?
Muse can help with online shopping, emails, trip planning, and generate AI podcasts, images, videos, and interactive webpages called artifacts.
What privacy concerns did the reviewer have?
Muse pulled detailed interests from Instagram's API data, which the reviewer couldn't see, and inferred her location from an Amazon shipping address, leaving her uneasy about letting it handle personal emails and credit card.
Did Meta respond to the privacy concerns?
The Verge reached out to Meta about the data Muse can access inside Instagram and Facebook but didn't immediately hear back.

Get the latest Large Language Models news every morning

For example, today's edition would include:

  • Dynatrace acquires Arize AI as observability shifts to actionSiliconANGLE AI · 2h ago
  • Shared base cuts 100 fine-tunes from 1.5 TB to 19.3 GBDaily Dose of Data Science · 2h ago
  • OpenAI agents hit RubyGems, undisclosed since May 12thSimon Willison's Weblog · 2h ago

AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. Q&As are published on this page for other readers too.

Related Articles

Next articleOpenAI launches ChatGPT for Financial Services with GPT-6 Astra