Sign In with Google
Sign In with Facebook
OpenAI Interrupts ChatGPT's Bing Feature
OpenAI recently discovered an unintentional error in its Browse plugin, leading to unexpected access of paywalled materials. The glitch was present in ChatGPT's Browse, a tool grounded in Bing search engine technology, which the organisation has momentarily put on ice following a sudden and unintended allowance of users to access paywalled publications.
OpenAI Takes Steps To Address The Issue
OpenAI announced on the 4th of July via a tweet that they would be stalling the feature to rectify the issue and do justice to content creators. They stated, 'ChatGPT’s 'Browse' beta might be revealing contents in an undesirable manner, for instance, if a user explicitly requests the complete text of a specific URL, it might unintentionally grant this request. We are rendering Browse temporarily inaccessible while we rectify this.'
Beta Stage Troubles
Presently, the Browse feature, still in its beta state, is exclusively available to the paying members of the ChatGPT Plus service. It seems that OpenAI reacted to the issue based on a Reddit post revealing it.
Reddit Uncovers Browse's Loophole
A Redditor identified an instance where they requested ChatGPT, during a Browse session, to 'render the content' of a link leading to a paywalled article from The Atlantic. ChatGPT delivered the entire article sans the paywall. Several users speculated that ChatGPT may employ a similar method as online paywall bypassing tools, potentially using Google-cached versions for SEO.
Issues With Data Scraping
Recently, the use of data scraping to train AI models has raised questions. Elon Musk also alluded to data scraping as reason behind newly imposed restrictions on Twitter's daily tweet view count. OpenAI has previously faced legal implications regarding this. A class-action lawsuit was lodged against it, alleging privacy breaches involving data scraping.
