OpenAI faces lawsuit for using 'stolen' personal data to train ChatGPT

OpenAI is now struggling with a class action lawsuit filed by the Clarkson Law Firm alleging that the company used "stolen" personal data without consent to train its language models, particularly ChatGPT and Dall-E.

The complaint, filed in Northern District Court in California, alleges that OpenAI collected 300 billion words from the Internet, including personally identifiable information of millions of Internet users, including children, without their informed consent or knowledge. The law firm also accuses OpenAI of conducting these activities covertly, without complying with the requirements of the law to register as a data broker.

OpenAI has already faced validation of its data collection practices, especially with respect to user consent and privacy protections. Until recently, users had no clear ways to opt out of using their conversations and personal information with OpenAI to learn and improve ChatGPT.

In fact, ChatGPT was temporarily banned in Italy due to concerns over insufficient user data protection, especially for minors, in accordance with the European Union's General Data Protection Regulation (GDPR). While this lawsuit addresses OpenAI's lack of transparency about its privacy policies for existing users, its primary focus is on the utilization of data collected from the web, which was never intended for ChatGPT.

The suit argues that OpenAI has profited from this data through significant investments from Microsoft and revenue generated from ChatGPT Plus subscriptions, all without compensating the original data sources.

The complaint outlines 15 counts, including violation of privacy, negligence for failing to protect personal data, and larceny by illegally obtaining massive amounts of personal data for training purposes.