Trusted AI is Governed AI - Taking Control of Your AI Data
Part 3 - How to see what AI knows About You
In Part 2, I introduced the Personal AI Data Map. It showed possible storage points during an AI interaction - but remember - not every interaction touches all of them.
And each LLM-enabled tool has its own version of this map. If you use ChatGPT, Gemini, Perplexity, and work systems, your footprint may be spread across them all.
How Your Data Interacts with LLMs
LLMs don’t magically absorb your information into their “brains.” Instead, your data may land in logs, session histories, or long-term memory features. Some of it may later be used to retrain future models.
The problem: few companies provide enough detail to know where, how long, or by whom your data is accessible.
What Do LLMs Actually “Know” About You?
There are two main kinds of knowledge:
Explicit Knowledge: What you provide directly (e.g., your name or résumé).
Inferred Knowledge: What the AI deduces based on patterns in your requests (e.g., frequent stock questions → you’re a trader).
Inferred knowledge is especially risky: you may never see or control it, yet it shapes your AI experience.
The Transparency Problem
Unlike browsers that at least show a history, most AI systems provide no audit trail. Deletion options vary, and users are often left guessing.
What You Can Do
Here are practical steps to gain visibility:
Create your AI inventory. List every LLM you use - personal and work. Assume that anything entered into a work system may belong to your employer.
Review terms and conditions. Dull? Yes. Necessary? Absolutely.
Query the LLM directly. Some systems will share what they “know” about you. Try prompts like:
“Please summarize everything you currently know or remember about me based on this conversation (and any memory features, if available). Be explicit about:
Information I’ve provided
Inferences you’ve made
Any stored memory beyond this chat
Assumptions you’re making
What will or will not carry over into future chats”
It won’t give you the whole picture, but it will illuminate the shadows.
Next
In Part 4, we’ll apply these ideas to a real-world case: what happens when an AI system shuts down - and why data portability does not equal experience portability.
How can I help you “Keep Pace” with AI and Intelligent Agents?
With the rapid evolution of AI, a strong AI governance function is mandatory. I offer a range of consulting, governance and architecture services designed to help organizations successfully achieve their AI Goals.
I also host a free weekly webinar every Sunday at 9:00 AM EST, where I break down the week’s most important AI news and articles—and discuss what they mean for you and your business.
The webinar streams LIVE on the following sites:
www.facebook.com/keeppace
www.twitter.com/keeppace
www.linkedin.com/in/keeppace
And the history (and playlists) of previous videos is available at:
www.youtube.com/keeppace