In this episode, Dario García de Viedma discusses Mythos, an AI model capable of identifying and exploiting software vulnerabilities autonomously. The conversation explores the risks of AI in cybersecurity, its impact on open source ecosystems, geopolitical dependencies in AI development, and regulatory challenges.

Darío published two articles (only available in Spanish) on Anthropic's Claude Mythos at El País (https://elpais.com/tecnologia/2026-05-03/claude-mythos-la-ia-frente-a-la-ciberdelincuencia-a-escala-y-los-limites-de-la-seguridad-nacional.html) and at Real Instituto Elcano (https://www.realinstitutoelcano.org/comentarios/mythos-automatizacion-y-concentracion-de-la-ciberseguridad/).

We discussed several developments that have taken place in the field of cybersecurity, namely:

- CrowdStrike's upload of a faulty content update, triggering a global IT outage affecting approximately 8.5 million Windows devices: https://www.bbc.com/news/articles/cpe3zgznwjno.

- Anthropic's Project Glasswing to distribute thousands of tokens across private actors: https://www.anthropic.com/glasswing.

- 70-90% of software applications depend on open source components: https://www.linuxfoundation.org/blog/blog/a-summary-of-census-ii-open-source-software-application-libraries-the-world-depends-on.

- Anthropic has already provided access to the Pentagon with access to Mythos: https://cryptobriefing.com/pentagon-pushes-ai-first-strategy-boosting-anthropic-mythos-prospects/.

- AI Security Institute report on Mythos: https://www.aisi.gov.uk/blog/our-evaluation-of-claude-mythos-previews-cyber-capabilities.

- The Pax Silica alliance: https://www.state.gov/pax-silica.

To learn more about Dr. Alba Ribera Martínez and her research, we invite you to visit her website: https://www.albariberamartinez.com.

She also writes, from time to time, in her newsletter, the DMA Agora, about the latest developments surrounding the European regulation: https://www.linkedin.com/build-relation/newsletter-follow?entityUrn=7344021393451184128.