I don’t know how far OpenAI has come in fixing these issues, but I’m hoping we can at least get some functionality, so:
Let’s get our web-reading feature back!
Here’s my proposed solution:
Part 1: remove it’s ability to click links and “browse the web”, just convert the browsing model into a basic “web-reader” It will only process specific links we provide, avoiding the previous issues with it circumventing pay walls.
Part 2: add a bit of text to the terms of use saying that you (the user) is responsible for the content in the links you provide.
I believe this could help solve the issue. What do you guys think?
Yeah there’s definitely the option of remaking the functionality yourself though the API or just using one of the many plugins people have made, I just don’t find them as good as the “original”