Dashboard D2 · Module 2

🤖 AI Paste Simulator

Pick a work paste, pick a tool, hit send — and watch where the text actually goes. Then sanitize the same prompt and see that the answer barely changes. Nothing here talks to any real AI; it all runs in your browser.

Retention/training behaviour shown per tier is accurate as of 2026-09-03 and vendor-dependent — defaults change, which is itself part of the lesson. On personal/consumer accounts at that date: ChatGPT Free/Plus trains on your chats unless you switch it off (Settings → Data Controls); Gemini consumer chats can be sampled for human review and kept up to 3 years; Claude consumer chats are retained up to 5 years if you allow training, ~30 days if you don't. Business/enterprise/API tiers of all three: not trained on by default. The rule that never changes: an unapproved or personal tool means data outside THG contracts and visibility.

What are you pasting?

Which tool / account?

Your prompt

red = sensitive specifics · green = safe placeholders

Where your paste goes

The answer you'd get back

🤖 With this prompt

🎯 Usefulness to you

Try this: P&L → public chatbot → watch every red step light up. Hit 🧼 Sanitize — same tool, and suddenly nothing sensitive leaves the building, while the answer is almost identical (you fill the real numbers in yourself). That's the whole of Module 2: structure stays, specifics go — or use the approved tool and paste the real thing.