Privacy and LLMs

LLMs like ChatGPT, Claude, Perplexity and other AI chatbots seem to be grabbing more personal information from innocent users today. Privacy seems to be lost for users and their personal information is at stake here.

This post discusses some ways in which data privacy can be restored to some extent.

Due to their conversational and invisible nature, it is very easy for users to pour their personal worries, medical queries, relationship woes and many other things and ask for suggestions to solve them. Just by being a silent listener and a non-judgemental machine, LLM’s offer more solace than a normal human being. In addition, LLMs also infer a lot of information from the questions being posed to them.

Hence, in addition to the huge amounts of data being shared, they understand more and draw a digital persona of us.

This results in huge amounts of personal information being gathered by them and a vast amount of data being stored as well. This information may be stored for any amount of time and used to train their models and know more about you (there are measures that can restrict this in ChatGPT by “Turning on temporary chat” or turning off “Improve the model for everyone” in ‘Settings’) but it is not accurately known if all the personal information will never be saved.

It is therefore best to not share any personal information at all. Treat the LLM as a stranger whom you will never share any personal information. It is in the best interest to only have a business relationship with any LLM be it ChatGPT, Perplexity or Claude.

So, in the best interest of privacy, it is good to follow these practices to safeguard our information better:

Here’s to a safe and private space when browsing the Internet!

This post is for alphabet ‘P’ for BlogchatterA2Z by @blogchatter