I can decide not to use Meta’s Muse. I can refuse to upload my data. I can opt out of training.
And yet, if people I communicate with let an AI agent ingest their inboxes, chats, files and histories, then it may ingest me too.
Decade of my emails sit in their inboxes. My messages are in their archives. My habits, preferences, relationships and half-baked opinions are embedded in other people’s data.
Take something completely mundane. I message a friend:
“Dinner at 8? Somewhere vegetarian. And please don’t invite Marco.”
That is not exactly a state secret.
But if my friend gives an AI agent access to years of messages, the system may now know that I’m eating less meat, that Marco and I are not currently enjoying a golden age, where I tend to eat, who I spend time with, and perhaps quite a bit more once it connects the dots.
I never told the AI any of this.
My friend did.
Now scale that up.
I email Alice. Her AI sees it.
I message Bob. His AI sees it.
Carla forwards something I wrote. Her AI sees that too.
Enough fragments distributed across other people’s inboxes could start to resemble a crowdsourced shadow inbox of mine.
And perhaps an even stranger one: My own inbox contains what I said. Other people’s inboxes may contain what they said about me, how they interpreted what I said, and how they reacted to it.
An AI able to connect those pieces could potentially know not only what I wrote, but what everyone else thought I meant and what they thought and did.
That is even more revealing than access to my own mailbox alone.
People could always forward an email, show someone a message, or gossip over dinner. But they usually did not ingest years of correspondence, cross-reference it with documents and calendars, infer patterns and maintain a searchable memory of my life.
We still talk about privacy as an individual choice: “Do you consent?”
But with AI agents, my privacy increasingly depends on what everyone around me consents to.
Everyone around me has pieces of me. AI makes putting the pieces together much easier.
At that point, whose data is it really?
And what exactly does my “opt out” mean if everyone else can opt me back in?
———
Originally posted on LinkedIn on October 2, 2026, the post by Gianluca Miscione discussed privacy concerns raised by non-users regarding Meta’s Muse AI agent.
Gianluca Miscione is a researcher exploring information systems across Europe, America, Asia, and Africa, focusing on health, urban planning, digital environments, organizational innovation, information infrastructures, qualitative and mixed methods, trust, and computer-mediated communication.
He is currently an academic visitor at Oxford University.
______________________
Bookmark Techparley.com for the most insightful technology news from the African continent.
Follow us on X/Twitter @Techparleynews, on Facebook at Techparley Africa, on LinkedIn at Techparley Africa, or on Instagram at Techparleynews
