Cyber news, tools & one smart career path.
5 min read
Quick Reality Check
A report from 404 Media found that OpenAI is reportedly using hundreds of contractors to read and review real ChatGPT prompts and responses as part of a program called “Project Lily.”
What happened:
“Project Lily” is reportedly a program in which contractors review, score, and critique ChatGPT’s answers to help improve the chatbot’s quality and behavior.
OpenAI says it uses a “Privacy Filter” to remove personal information before prompts reach reviewers. They may not see your username, but they may still see personal information you typed into the chat and, in some cases, a summary of how you have used the chatbot before.
Why it matters:
AI chatbots are not private diaries. If a conversation is selected for review, personal details, health questions, legal issues, or family concerns you typed into ChatGPT may be seen and scored by human contractors, even after a filter removes obvious identifiers.
The risk is that your prompts, the personal information you typed directly into the chat, and summaries of your past usage may still be visible. That means sensitive details can end up in front of strangers.
Remember that convenience and privacy are not the same thing. The more personal information you give an AI tool, the more careful you should be about its settings and policies
Read more here
60-Second Protection Fix
How to Opt Out of AI Chatbot Training
-
ChatGPT: click your profile icon, go to Settings > Data Controls, and turn off Improve the model for everyone. This setting is on by default for Free, Plus, and Pro accounts. Turning it off stops new chats from being used to improve the models, although it does not prevent every form of authorized access for security, support, legal, or abuse-related reasons.
-
Perplexity: Go to Settings > Preferences > AI data retention and turn it off. This prevents data collected after you opt out from being used to train AI models, but it does not erase information that was already collected.
-
Claude: Click your name, go to Settings > Privacy, and turn off Help Improve our AI Models. Anthropic says this stops new chats and coding sessions from being used in future training runs.
The bottom line: If you would not say it out loud in a coffee shop full of strangers, do not type it into AI
Cybersecurity Awareness Month is right around the corner…
I would love to know what you'd like to see and learn. What cybersecurity topics, scams, privacy concerns, or online safety questions should I cover?
What You Missed This Week
YOUR I.D. IS PROBABLY ON THE DARK WEB NOW.
Here’s what you can do ➡️
AI is making fake job offers look real.
Here is the one check that saves you.
Your LG smart TV is listening to you
Here is what happened
Let’s keep building together!
Stay protected,
Cybersecurity Girl
Know someone who’d enjoy this? Pass it along and have them sign-up here! And if you have thoughts or feedback, just hit reply, I’d love to hear from you.
Responses