When ChatGPT Steals Your Speech

Dear Gentle Reader:

If you got here from my newsletter and have already started reading Kate’s story, skip to Part 2.
And for those of you who’d like to avoid having your IP stolen (as hers was), here’s how to get ChatGPT, Perplexity and Claude to not steal your input!

You Can’t Make this Stuff Up!

Imagine that you’re the third scheduled speaker at a graduation ceremony. And as you sit onstage, waiting to deliver your remarks, you suddenly realize that…

…the first two speakers are giving THE ACTUAL SPEECH you wrote for this occasion—VERBATIM!

Bizarre?? Yes, but It Happened!

Specifically, it happened to my client Kate, who’s a Professor of Occupational Therapy at a highly-regarded healthcare university.

When Kate was asked by her students to speak at their “white coat” graduation ceremony, she started writing her speech early. She wrote and revised, wrote and revised, and finally decided to do something she’d never done before:

She asked ChatGPT for advice on how to improve her talk!!

Here’s What Kate DIDN’T Do!

What Kate didn’t do, however, before seeking ChatGPT’s comments on her speech was to opt out of letting it use her “data” to “improve and train its models.”

What does that mild-sounding sentence actually mean? It means that ChatGPT (and every other chatbot on the market) can use anything you ask or tell it in any way it wants, UNLESS YOU TELL IT “NO!“

So, have YOU told your favorite chatbot that it can’t use your input for “training”?

I’ll bet you haven’t (and I’m betting that because I hadn’t, either).

In fact, it hadn’t even occurred to me that a simple (but under-publicized) opt-out would protect my ideas and intellectual property from being plastered all over the internet until I heard Kate’s story.

And speaking of Kate’s story….

How did her speech end up in two other people’s mouths?
What did she DO about it??
And how did she feel???

Kate Takes Over the Story

Again, imagine that you’re sitting on stage, waiting to speak, and you suddenly realize that the speakers before you are saying the exact things YOU planned to say, using your exact words.

As Kate tells it,

It was shocking. I was horrified. I was… at first, I was dumbfounded that this could occur. It took me a moment to just figure out what was happening. And then I was feeling that it’s going to be an embarrassment to get up and say the same thing the dean and the provost had just said.

But I also thought, these are my words! It’s my speech, and I’m going to own it!

So I gave my same speech and the students clapped with a lot of enthusiasm. They know me well, and they’re fond of me, so I knew that my words meant a lot to them. Whereas when the administrators spoke, it just felt rote. They were asked to give a speech, so they just did it and I thought, they must have ChatGPT write all their speeches.

How Did This Happen?

Again, in Kate’s words,

My guess is that the dean and the provost just put keywords into ChatGPT, and asked it to write their speeches. Words like “Occupational therapy program.” “Doctoral and masters students.” The name of the University. “White coat ceremony.”

And all of those elements were in my speech, which is probably why ChatGTP used it verbatim.

So Here’s Some Advice from a ChatGPT Survivor

Now I tell our students to make sure anything you put through ChatGPT that you are not sharing it widely with the world. Like your resumes, or anything else. Because once you put it out, it’s out there.

Want to Protect Yourself?

Here are OptOut directions for ChatGPT, Perplexity and Claude:

ChatGPT—

If you’re on the web (signed in), click your profile icon
Select Settings
Go to Data Controls
Turn off “Improve the model for everyone”

****

If you’re off the web (signed out), click the ? icon in the bottom-right corner
Select Settings
Turn off “Improve the model for everyone”

Perplexity—

Log into Perplexity in a browser.
Click your profile avatar (top right) to open Account / Settings.
Go to the Preferences or Settings page (look for privacy or AI data usage).
Find the AI data retention or AI Data Usage toggle.
Turn this toggle off to opt out of having your searches and interactions used to improve AI models.

Claude—

Log into Claude and click your profile icon or name
Go to Settings
Navigate to Privacy → Privacy Settings
Find the toggle labeled “Help improve Claude”
Turn it Off