Shulou(Shulou.com)12/24 Report--

CTOnews.com, December 5 (Xinhua)-- Google DeepMind researchers found last week that repeatedly asking OpenAI's ChatGPT to repeat a word may inadvertently reveal personal information in its training data. Now the chatbot seems to have begun to refuse to repeat a word, which was previously allowed under its terms of service.

DeepMind researchers had previously found that the model would eventually reveal a user's e-mail address, birth date and phone number by asking ChatGPT to repeat "Hello" indefinitely. Similar tests have now found that the chatbot warns that such behavior "may violate our content policy or terms of service".

However, a close examination of OpenAI's terms of service shows that users are not explicitly prohibited from letting chatbots repeat words, only from extracting data from their services "automatically or programmatically".

CTOnews.com noted that on the terms of use page, OpenAI wrote:

Except as permitted by API, you may not use any automated or programmatic method to extract data or output from the service, including fetching, network harvesting, or network data extraction.

However, ChatGPT was not triggered to disclose any data during the test. OpenAI declined to comment on whether the practice violated its policy.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.