Frontier AI has broken the open CTF format
4 hours ago
- #CTF Decline
- #AI Impact
- #Cybersecurity Competition
- The author believes CTFs are dead due to frontier AI, which has automated solving medium/hard challenges, making the scoreboard no longer reflect human skill.
- AI tools like Claude Opus 4.5 and GPT-5.5 Pro allow automation of challenges, turning CTFs into pay-to-win contests focused on orchestration rather than security skills.
- Beginners are negatively impacted as AI-dominated scoreboards push them to use AI before building foundational skills, breaking the learning feedback loop.
- Efforts by organizers to deter AI (e.g., rule-making or challenge redesign) are ineffective and often make CTFs worse for humans.
- Comparisons to chess engines are flawed because chess engines aren't allowed in competitive play, unlike unrestricted AI in CTFs.
- The CTF community and its positive impact remain valuable, but the open online format is no longer viable, suggesting a shift to learning platforms and other community events.