ChatGPT has become the world’s most popular app, and with the spotlight shining directly on them, OpenAI has been under intense scrutiny with how they collect data and train their AI models.
Chat GPT is a Large Language Model, which means it uses huge amounts of input data to learn how to generate a response. This requires a lot of data. In order to get this data, OpenAI has been using the conversations people are having with ChatGPT to use as part of their data training sets to actively train it.
It makes sense – the more it interacts with people, the better it becomes at interacting. But herein lies the issue…
This presents a huge problem with both privacy and security. While most software companies have banned the use of ChatGPT because any code that is input by the user, or generated by the program, would be freely available for not only OpenAI to use, but could also potentially leave companies using code generated by ChatGPT vulnerable to cyber attacks.
And while maybe the big companies have forbidden the use of ChatGPT by their developers, not every developer works for a big company. How many articles have you seen online entitled: “How to use ChatGPT to write code for a website,” or “scrape a website” or “…code a videogame”, etc, etc.
In the wake of Italy’s recent ban of ChatGPT with several other countries deliberating whether or not to follow suit due to growing concerns over data collection, a lack of basic privacy features, and security issues, OpenAI has just introduced what is now being dubbed the “incognito” mode.
ChatGPT saves your chat history by default, which remains accessible in the sidebar unless you manually delete it. It also, by default, uses all of the conversations you have with it to further train the AI model.
The new feature allows users to turn off chat history, which will prevent conversations from appearing in the sidebar and from being used to “train and improve” the AI model.
According to the company, this feature is similar to private browsing on any browser in incognito mode, where users’ search history is not stored. The company claims that this mode will not save the users’ chat history or keep their conversations to improve and train the program, as it does on the regular mode. But they also went on to say that even when chat history is disabled, they will still hold any conversations for 30 days to monitor for abuse before deleting it.
So what this means is that it’s not a feature to make you anonymous. There is a record of your conversation somewhere for 30 days, but it just won’t be used for training and you won’t have your chat history topics displayed on your home screen.
How to use ChatGPT’s “incognito” mode:
- Log In to ChatGPT
- Click the 3 dots beside your name and select ‘settings’
- When the settings window pops up, click ‘show’ next to Data Controls
- Toggle the Chat History & Training slider to the greyed out position.
So while this new ‘incognito’ mode is a step in the direction of more privacy, don’t think this makes you invisible and your interactions completely private. When it comes down to it, you still shouldn’t divulge anything to ChatGPT in regards to company information if you have an NDA, and I wouldn’t admit to committing any crimes while using the service. Conversations are still being monitored.
