Hacker News Reader

This is pretty fun and a clever way to collect wild prompts and analytics about how well their AI safety suite is doing :)

Would be nice if there was an open source version of this, where the data was published for the public to learn from

It's something we're actively looking into!