OpenAI's ChatGPT Introduces Incognito Mode for User Privacy
OpenAI has announced new features for their chatbot, ChatGPT, focused on addressing user privacy concerns. The San Francisco-based startup is introducing an "incognito mode" that will not store user conversation history or use it for improving artificial intelligence. ChatGPT Business, a subscription service with increased data controls, will also be launched.
The move comes in response to growing scrutiny over how chatbots like ChatGPT manage users' data, which is used to "train" AI. Italy banned ChatGPT last month over potential privacy violations and called on OpenAI to provide consumers with tools to object to data processing. France and Spain also began investigations on the service.
OpenAI's Chief Technology Officer, Mira Murati, stated that the company is working to assure regulators and is compliant with European privacy laws. The new features showcased were not developed in response to the Italy ban, but rather a prioritisation of user privacy the company has been focused on for months.
New features include the ability to turn off "Chat History & Training" in user settings and export their data. OpenAI will still retain conversations for 30 days to detect any potential misuse, after which they will be permanently erased. The upcoming ChatGPT Business subscription will not use conversations for AI model training by default.
Nicholas Turley, OpenAI's Product Officer, compared the new incognito mode to an internet browser's private browsing feature and noted that the company is committed to putting users "in the driver's seat" regarding data collection. While user information helps OpenAI improve its software and reduce political bias, among other issues, Murati acknowledges the challenges the company still faces.
OpenAI has partnered with Microsoft, which has also invested in the company, to offer ChatGPT to businesses. The new business subscription launched is expected to appeal to the cloud provider's existing customers. The goal of OpenAI is to create AI models that prioritise privacy while being super aligned with users' preferences.
