AI Threatens to Sideline Champion Ethical Hackers, Pwn2Own Winner Warns

BBC Business reported Tuesday that one of the world’s top competitive hackers believes AI may soon make elite human talent redundant in cybersecurity research.

Valentina Palmiotti, known in hacking circles as Chompie, was the standout individual performer at this year’s Pwn2Own competition in Berlin. The annual contest, run by the Zero Day Initiative, pays ethical hackers to uncover previously unknown software vulnerabilities before criminal actors can exploit them.

Chompie’s Warning on Ethical Hacking AI

Palmiotti walked away with $70,000 in prize money across two successful hacks. One targeted an Nvidia-linked system, earning her $20,000. A second, harder challenge against a Linux-based system netted $50,000 after an all-night research session she calls “zombie hacker mode.” She described the process as hours of caffeine-fueled testing, often without sleep. She acknowledged the lifestyle was far from healthy but said it remained essential to win.

Despite her success, Palmiotti told the BBC she entered this year’s contest partly because she feared it might be her last realistic opportunity. She currently uses AI tools, including Claude Code, to accelerate her work at IBM X-Force and in competition settings. For now, she said, skilled hackers occupy a productive middle ground where AI is a useful assistant.

Background: Why Claude Mythos Has the Industry Worried

That window may be closing. Anthropic’s Claude Mythos model has drawn significant attention across the cybersecurity community. The company claims the system independently identified more than 1,600 vulnerabilities across hundreds of software programs. Anthropic has restricted access to select governments and specialist security institutions, citing the model’s potential for misuse. Palmiotti pointed to Mythos and OpenAI’s GPT-5.5 Cyber as the tools most likely to raise the competitive bar beyond human reach.

Her view: future demand will exist only for the absolute best human researchers. She cited Taiwanese hacker Orange Tsai as an example. His team won $375,000 at Pwn2Own Berlin by mapping exceptionally complex attack pathways.

Orange Tsai Sees a Different Future

Tsai offered a more measured outlook. He told the BBC that AI functions as a powerful research accelerator rather than a replacement. He noted that the need to sleep limits how many ideas he can test manually, and that AI helps fill that gap. He remained hopeful that human intuition and creativity would continue to surface vulnerabilities that automated systems miss.

On the threat side, researchers note criminals are also deploying AI to speed attacks and develop new intrusion methods, though most real-world breaches still rely on simpler, well-established techniques like phishing.

Read Next: AI Is Reshaping Cybersecurity Faster Than Most Firms Can Respond