ChatGPT adds new privacy feature after worries over how it is collecting data

New tool added after Italy banned service

Artificial Intelligence-Audits (Copyright 2023 The Associated Press. All rights reserved)

ChatGPT’s creators have aded the option to turn off “chat history” within the system, allowing people to better control how their data is used.

The new tool comes amid concern about what data ChatGPT is gathering on its users, and how it is being fed back into the model that powers it.

Given that ChatGPT’s creators, OpenAI, refine the system so that future versions may be improved by the data gathered as people use it today. As such, data collected could potentially appear in the system’s future answers.

Now users will be able to switch off history, so that data will not be gathered by ChatGPT. Conversations will still be kept for 30 days, however, but the company promises only to review them “when needed to monitor for abuse”, after which it says it will permanently delete them.

“Conversations that are started when chat history is disabled won’t be used to train and improve our models, and won’t appear in the history sidebar,” OpenAI said in its announcement.

The new tool is rolling out to all users “starting today”. It is found in ChatGPT’s settings and can be switched at any time.

OpenAI only said that the feature was intended to be “an easier way to manage your data than our existing opt-out process”. It did not give any further information about why it had opted to roll it out.

But it comes amid loud and increasing criticism of the privacy policies of ChatGPT and other similar AI systems. Users have voiced concern about what data might be collected by those systems, and how it is regulated.

The most significant of those concerns came from Italian officials, who last month banned the tool in the country. It expressed concerns over the legal basis that OpenAI uses to collect personal information, and how that information might be used to train the system.

It also comes after ChatGPT’s “history” feature went awry just a few days before Italy made its decision. In late March, OpenAI said that it had experienced a “significant issue” that meant users could see information about other people’s chats.