ConeheadAI · leftover lesson

SNK3-093 · Curriculum How-to

AI inside Docs/Sheets is still a send risk

L0–L2 · Curious–Capable (bootstrap — system rebalances) · ~4m

toolsprivacyworkplacetoolsINF · J5 · S3C1.3C1.5

Look for this insight

Workspace add-ons put models where you already work — powerful, and easy to leak client data if share/send is still human-owned only in theory.

Lesson hook

Grok showing up inside Google Workspace (Docs, Sheets, Slides) means the model now sits right next to your real files. That's a great place for draft help — and a risky one for autopilot sharing. Before you paste a client sheet or hit anything like send/share/export: strip out secrets, confirm who can see the doc, and keep a human owning any external send. In-app AI is not a privacy boundary.

Do this now

Open your notes. For your last AI-in-docs use (or a hypothetical one), list: (1) the data classes in the file (public / internal / client) (2) who can open it (3) whether the model's output could leave your tenant. You're done when you've written the one rule you'll enforce before any share/export, in 15 words or less.

Completion check: writes_share_export_rule

Open snack link →

Staff review Sign in

Related by competency

All snacks · Take Question 1 · Team training