Plaintiffs in a class action lawsuit alleging OpenAI violated privacy rights for training data dropped their case against the company. Court documents showed the case was dismissed without prejudice, and the plaintiffs can choose to refile. 

The lawsuit, first filed in June this year in the Northern District of California, alleged OpenAI’s web scraper “violated property rights and privacy rights of all individuals whose personal information was scraped and then incorporated through misappropriation into [OpenAI’s] products.” The lawsuit did not name the plaintiffs, who were identified with initials. The Clarkson Law Firm filed the class action suit on their behalf.

OpenAI, like other generative AI companies, scrapes publicly available data from the internet to help train large language models.

Questions about how generative AI companies like OpenAI ingest and use publicly available data to train their models have led to several lawsuits. Most cases revolve around the thorny issue of copyright rather than privacy rights. Comedian Sarah Silverman and authors Christopher Golden and Richard Kadrey sued OpenAI and Meta for allegedly infringing copyright to train GPT-4 and Llama 2.

In July, the Federal Trade Commission opened an investigation into OpenAI for possible consumer harm through data collection and the publication of false information. 

In August, OpenAI said website owners can now block its web crawler. Several websites, including news publishers like The New York Times, have blocked OpenAI from scraping its data.