# Cybersecurity Looks Like Proof of Work Now

- slug: cybersecurity-looks-like-proof-of-work-now
- date: 2026-04-14
- category: Artificial Intelligence

Anthropic says its Mythos model lets defenders outbid attackers on compute. An independent AI security company tested the same vulnerabilities with an $0.11/M model — and the cheap one found them too.

---