Anthropic's Mythos Model Cracks Post-Quantum Crypto Flaws in 60 Hours for $100k
The Decoder · rss · 2026-07-29
Anthropic announced that its Claude Mythos Preview model successfully discovered vulnerabilities in core cryptographic algorithms securing the internet.
- Key Discovery: The model found a better attack on HAWAK, a post-quantum signature scheme previously reviewed by human experts for over two years.
- High Efficiency: Uncovering the flaw took only 60 hours at an API cost of roughly $100,000.
- Industry Impact: While the findings don't affect systems in use today, they demonstrate AI's potential to challenge the foundational assumptions behind internet security.
More from Safety
- MCP Toolbox Introduces URL Parameter Binding to Keep Sensitive Data From LLMs — rseroter · 2026-07-29
- Noahpinion says AI job loss and AGI takeover look unlikely, but bioterrorism risk is high — ZeroStateReflex · 2026-07-29
- Why the AI Pause Movement Struggles: Radicals Drain Valuable Social Capital — airkatakana · 2026-07-29
- Open models helped find six NGINX vulnerabilities across five CVEs — cyb3rops · 2026-07-29
- Rapid7 says a SmartConsole bypass can hand attackers full admin access — cyb3rops · 2026-07-29
- LLM multi-agent workflow finds open-source 0-days across Nextcloud, Grafana and more — cyb3rops · 2026-07-29