How To Stay Private On ChatGPT
Ever had the eerie feeling that you're oversharing with ChatGPT? That your AI companion doesn't need to know the Q&A on your latest colonoscopy (the third one this year)? Or maybe you're hesitant about divulging business secrets to an AI? In this guide, we’ll show you how to put on your virtual sunglasses and use ChatGPT with enhanced privacy.
Section 1: Turning off Chat History
OpenAI is not particularly interested in reading your conversations. However, it does log your chat data to train and improve its AI models (aka reinforcement learning). Recognizing the need for an extra layer of privacy, ChatGPT rolled out a feature in April that lets you turn off your chat history. This means your chats are permanently deleted after 30 days, leaving no digital breadcrumbs.
Just follow these simple steps to turn off chat history:
- Click on your username in the bottom left of the screen.
- Navigate to "Settings."
- From there, select "Chat History and Data."
- Here, you’ll find an option to turn off your chat history. Give it a click, and voila!
Should you not mind ChatGPT learning from your conversations, feel free to keep the chat history on - it does contribute to the continuous improvement of the model.
Section 2: Dealing with Work Data
While toggling off chat history addresses some privacy concerns, it may not suffice for the ‘ol boss. Several organizations, such as JP Morgan and Samsung, have prohibited employees from using ChatGPT due to data privacy risks.
In response, ChatGPT announced plans for a business subscription, which would offer enterprises complete control over their data. However, this isn’t live yet. So, where now?
Surprisingly, the pool of AI chatbots that don't record your conversations is quite shallow (every company wants to improve their models from your data!).
Nevertheless, a few options stand out from our research:
- Using GPT-4 via Azure OpenAI: Microsoft offers robust safety measures to protect your data, ensuring all customer data stays within Azure OpenAI service without being sent to OpenAI.
- Exploring specific-use AI platforms: Salesforce's AI Cloud, for instance, targets sales and marketing and has explicit policies ensuring that your company's sensitive data remains secure.
- Use AI locally: Opt for versions of AI models that run locally on your own servers. This puts you in the driver's seat regarding data control.
Conclusion
And there you have it - three ways to make your AI use secure, allowing you to chat freely without the digital overhang. Bada bing, bada boom! Your AI conversations are now on lockdown!
How else are you looking to use AI at work? Let us know the types of tutorials you want to see from us next here!