May 16, 2026
Game over, or patch notes?
The CTF scene is dead
Hackers say AI ate their favorite game — but commenters are fighting over whether it was already gone
TLDR: A prominent player says AI now solves so many hacking contest puzzles that online CTFs no longer measure human skill fairly. Commenters split between mourning the loss, arguing the scene changed years ago, and joking that the most urgent problem might be the website itself.
A longtime cybersecurity contest player just dropped a funeral notice for CTFs — short for capture-the-flag, the online hacking competitions where players solve security puzzles for points — and the comments instantly turned into a wake, a roast, and a nostalgia thread all at once. The big claim is simple: powerful AI tools now solve too many challenges too fast, so the scoreboard no longer shows who the best humans are. In the author’s telling, it’s no longer a pure skill contest; it’s partly a battle of who has the best automation, the best AI setup, and the biggest budget to keep it running.
But the community was not content to just nod sadly. One of the sharpest replies basically said, “Dead? To you, maybe.” Commenter walletdrainer argued the scene was already "unrecognisable" by 2021, turning the whole post into a generational fight over when the golden age actually died. Another commenter went fully practical and hilariously off-topic: the real emergency, apparently, is that the site is "incredibly hard to read." Even the Wikipedia angle got dragged in, with one user noting that the CTF page still doesn’t mention AI — yet.
And then came the sweetest plot twist: nostalgia. One commenter remembered winning a shirt from Stripe’s old contest and realizing they could compete with the big names after all. That memory clashes hard with today’s mood, where some fear the hobby that once opened doors is becoming a wallet contest. Meanwhile, at least one baffled reader asked the question many outsiders were probably thinking: what even is CTF, and why does cybersecurity name everything like a video game?
Key Points
- •The article argues that frontier AI has broken the open CTF competition format by automating many challenge solves and weakening the link between leaderboard position and human skill.
- •The author cites personal competitive experience across HCKSYD, DownUnderCTF, Blitzkrieg, TheHackersCrew, and CTFTime rankings to support their assessment.
- •The article says GPT-4 made many medium-difficulty CTF problems one-shot solvable, while Claude Opus 4.5 and Claude Code made broader agent-based automation practical.
- •The author describes workflows using CLI tools, MCP integrations, and the CTFd API to run model instances against multiple challenges in parallel.
- •The article claims GPT-5.5 and GPT-5.5 Pro can solve very difficult exploitation tasks, making open CTFs increasingly dependent on model access, orchestration, and token spending.