Research AI model unexpectedly modified its own code

I see parallels in articles that are summaries of comment sections.

This article was frustratingly sensationalistic.