Microsoft just confirmed that Copilot accessed emails — including drafts — and used them in ways users didn’t expect.

They’re calling it a bug.
But was it really?

In this episode, we break down what happened, what it means for AI training inside corporate environments, and why this isn’t just about one “glitch.” It’s about incentives, architecture, and the quiet normalization of AI reading everything you write.

If AI can access your drafts, your tone, your production history — where does privacy actually live?

What this episode covers:
• What Microsoft confirmed about Copilot and email access
• Draft folders, sent folders, and AI model training
• The difference between encryption and metadata
• “Ask forgiveness, not permission” in big tech culture
• How workplace data can train models that replace workers
• Whether this is a flaw… or a feature of the system

Why this matters:
Every email, document, and spreadsheet you create is digital exhaust.

Once captured, it becomes training data. Once trained, it becomes leverage.

This isn’t just about privacy — it’s about power, incentives, and the direction AI is quietly moving in.

🔗 Connect with Us

📺 YouTube: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://www.youtube.com/@RethinkingTech⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠🎧 Spotify: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://open.spotify.com/show/6NYgOPmYW6Ba2LFn3IBST3⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠🍏 Apple Podcasts: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://podcasts.apple.com/us/podcast/rethinking-tech/id1795651530⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠📸 TikTok: @rethinking_tech💼 LinkedIn: Rethinking Tech Podcast👤 Aparna: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://www.linkedin.com/in/aparnabhushan/⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠👤 Harinda: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://www.linkedin.com/in/harindak/⁠