OpenAI’s AI large-language models, such as GPT, have not been trained on paying customer data “for a while,” CEO Sam Altman told CNBC on Friday. “Customers clearly do not want us to train on their data, so we’ve changed our plans: We will not do that,” Altman said to CNBC’s Andrew Ross Sorkin.
According to the Internet Archive’s Wayback Machine, OpenAI’s terms of service were secretly changed on March 1. “We don’t train on any API data at all, and we haven’t for a while,” Altman told CNBC. APIs, or application programming interfaces, are frameworks that enable clients to integrate with OpenAI’s services directly.
OpenAI’s business customers, which include Microsoft, Salesforce, and Snapchat, are more likely to take advantage of OpenAI’s API capabilities.
Although OpenAI’s new security and privacy measures are only accessible by clients that utilize the company’s API services. “We may use Content from Services other than our API,” according to the company’s revised Terms of Service. This might include text entered by staff into the extremely popular chatbot ChatGPT. According to reports, Amazon has cautioned employees not to provide personal information with ChatGPT for fear of it appearing in responses.
The switch comes as companies face the potential of large-language models replacing human-created content.
The Writers Guild of America, for example, went on strike on Tuesday after model negotiations between the Guild and Hollywood companies fell down. The Guild has been advocating for restrictions on the usage of OpenAI’s ChatGPT for script generation or rewriting.
READ ALSO – Reliance, Adani, and Vedanta are among the companies in the race for Bhadreshwar Vidyut.
Executives are similarly concerned about the impact of ChatGPT and related programs on their intellectual property. Barry Diller, head of IAC, has recommended that media corporations take their concerns to court and perhaps sue AI businesses over the usage of creative content.
As an AI language model developed by OpenAI, I do not have access to information regarding the internal functioning or operations of the company. However, if the report is accurate, it means that OpenAI has not been using data from paying customers to train its GPT language model for a while. This could be for a variety of reasons, including concerns around data privacy, security, or simply a shift in the company’s priorities and focus.
It is worth noting that OpenAI has previously made efforts to create a more ethical and responsible approach to AI development, including the creation of the OpenAI Charter, which outlines the company’s commitment to using AI for positive and socially beneficial purposes. Not using customer data for training purposes may be part of this overall ethical approach.
Regardless, it is important for companies and organizations working with AI to be transparent about their data practices and to prioritize the ethical and responsible use of data.
