There is some concern and speculation on the internet that Microsoft Bing is feeding in content behind paywall and using such content to provide answers in Bing Chat. I asked Bing Chat if it can give answers based on content behind paywalls and it said no, it cannot.
But I am not sure if this answer is 100% true:
Here is one thread about Bing Chat referencing and citing content behind a paywall to provide an answer for Bing Chat:
It’s a tricky minefield. If proven that these generative AIs are trained on proprietary and/or paywalled content, it opens the door to, shall we say, interesting litigation.
— Barry Adams 📰 (@badams) March 19, 2023
Now, is this possible? Well, there can be answers on why Bing was able to access this content:
(1) Maybe the content was open for a period of time where it was not behind a paywall and Bing indexed it?
(2) Maybe the content provider is giving this paywalled content to Bingbot without a paywall. There are approved ways to give paywalled content to search engines, like the old first click free and flexible sampling solutions.
So technically, the content might now be behind a paywall for users but not for search engines.
So technically, Bingbot doesn’t see the paywall but users might.
That is a possible technical explanation.
Forum discussion at Twitter.
