A big fraction of the comments on this thread are about the impact of cheating on competitive games. It's important to understand that automating CTF challenges isn't usually cheating. It's normally part of CTF culture. The better teams have toolboxes ready to shred the early challenges; it's not a level playing field and was never intended to be.
(The author of the piece understands this; I think they're broadly right, though I think these games will find other ways to incentivize participation without the now-meaningless leaderboards.)
This is already addressed in the blog post about the fast that frontier LLMs have moved to being able to solve the kind of problem you'd expect a talented amateur or mid-level pro to do (aka top level CTF problems)