This Is The Reason OpenAI Refuses To Use Customer Data Again To Train Big Language Models

ChatGPT, do not use customer data for the coach. (photo: dock.pexels)

JAKARTA - OpenAI CEO Sam Altman confirmed that he will no longer use Application Programting Interface or API customer data, to train his Big Language Model (LLM).API is a framework that allows customers to connect directly to OpenAI software.

"The subscription clearly wants us not to train their data, so we have changed our plans. We will not do that," Altman told CNBC International, quoted Monday, May 8.

The company updated its Service Provisions for the new rules in early March but without announcement.

If the user operates the ChatGPT directly, the data will still be used for training, except for using incognito mode.

"We haven't trained any API data at all, we haven't done that for a while," said Altman.

Currently, the company's portfolio includes Microsoft, Salesforce, and Snapchat, which are more likely to take advantage of OpenAI's API capabilities.

However, OpenAI's new privacy and data protection only applies to customers who use company API services.

Seeing OpenAI Usage Requirements, companies can use Content from Services other than their APIs, including text that employees enter into ChatGPT chatbots.

Samsung, recently banned its employees from using chatbots such as ChatGPT due to security leaks.

A similar move was also taken by Amazon warning employees not to share classified information with ChatGPT for fear of appearing as an answer.

Altman's statement came at a time when OpenAI reportedly doubled its losses to around US$540 million (Rp7.9 trillion) in 2022.

The report from The Information, quoted from Indian Express, said the losses came from the development of ChatGPT and employed several executives from Google. The information was shared by three people who know the company's finances.