OpenAI temporarily disabled the browsing feature of ChatGPT, a Bing-based search engine, due to a bug that allowed users to bypass paid content. The decision to suspend the service was announced in a tweet on July 4, with OpenAI expressing the intention to address the issues and ensure fairness for content owners. The browsing feature, which is currently in beta and accessible to ChatGPT Plus subscribers, had encountered unintended behavior, such as inadvertently fulfilling requests for the full text of URLs, prompting the need for a fix.
The action taken by OpenAI was prompted by a Reddit post that showcased screenshots of a browsing session where a user requested the text of a paid article from The Atlantic. This post gained significant attention, with speculations emerging that ChatGPT may employ similar mechanisms to online paywall removers, potentially accessing cached versions of articles. Users pointed out that the AI model might overlook paywall codes, allowing access to content that would otherwise require a subscription or login. The issue of data scraping has been a growing concern, leading to li mitigations imposed by platforms like Twitter, which cited data scraping as the reason for a new daily tweet reading limit.
OpenAI has faced legal challenges related to data scraping in the past, and the creators of ChatGPT were recently hit with a class action lawsuit, alleging the unauthorized collection of private user information from the internet. These incidents highlight the importance of addressing data scraping concerns and ensuring appropriate access to paid content while maintaining user privacy and adhering to legal requirements.
