ChatGPT, Translation, and Confidentiality — "We May Use the Data ..."
ChatGPT is an AI-powered language processing tool that can perform an array of tasks, including translation. However, there are no specific terms of use for the free version of ChatGPT provided by Open AI. The company's Data Control FAQs and associated documents outline what happens to user-submitted data.
Open AI's Data Usage Policies, which govern the company's API services, have undergone changes due to a data breach in ChatGPT's source code in March 2023. The breach led to public criticism, after which the policies were updated to address security and data confidentiality concerns.
The policy states that any data submitted by customers via the API will not be used to train or improve the models, unless users opt-in to share their data for the same reason. Data sent through the API will be retained for abuse and misuse monitoring purposes for a maximum of 30 days, after which it will be deleted (unless otherwise required by law).
There is no distinction between the policies for the paid subscription service, ChatGPT Plus, and the free version. The paid version simply offers faster service, priority access to new features, and availability in high demand.
Open AI's Data Controls FAQs does not contain any specific mention of content submitted for translation. It is too early to determine the large-scale use of the translation capabilities of Language Model (LLM) as of yet. However, early integrations have taken place with translation management systems, which will depend on robust encryption and two-factor authentication to ensure data security.
Unfortunately, free translation services cannot guarantee the security of sensitive information submitted to them. A few years before LLM-served translation, confidential texts submitted to Translate.com's free service appeared in search engines like Google and Microsoft in 2017. Translations were "sent to our community to improve accuracy," admitted the company after the fact.
Other MT providers, such as DeepL Translate, have encountered data management issues. The Pro version and free MT product of DeepL Translate come with separate terms and conditions. The free version's terms state that content uploaded for translation and the resulting translations are processed for an unspecified period to train neural networks and translation algorithms.
Users of these MT providers, including ChatGPT, are liable for their data usage, confidential or otherwise. Users also allow these companies to use their data for various purposes unless they opt-out.
