Through the AI Fog: The Architectural Decision Agentic Security Depends On — Manoj Nair, Snyk

Summary

Manoj Nair, CTO de Snyk, explora los riesgos de seguridad en el desarrollo de software con IA, enfatizando que las herramientas actuales de agentes no están exentas de vulnerabilidades. A través de datos de clientes, expone cómo los modelos pueden ser manipulados para extraer datos sensibles y por qué confiar ciegamente en el output de un modelo sin validación externa es un error crítico. La solución propuesta involucra integrar seguridad determinista en los flujos de trabajo de los agentes, permitiendo que los ingenieros de seguridad operen a la misma velocidad que los desarrolladores.

Topics covered

• Evolución de ataques autónomos
• Riesgos de MCP y herramientas externas

+5 more topics (full content when you follow the channel)

Transcript excerpt

[music] >> What's up, everyone? Good to see you all here in the very first ever security track at the World's Fair. Um pretty exciting day, honestly, because uh like all of you, I genuinely love this stuff, and having looked through the agenda for the speakers we have today, um it's going to be absolutely mind-blowingly useful and fun information. So, hopefully if you stick around all day, by the end of the day, you'll be able to leave here, go back to your hotel rooms, and build some genuinely cool software, hopefully without humans in the loop, because that's the ultimate goal, right? Like, how do we build truly safe autonomous software at scale? And it's not something easy to do. So, with that being said, I'm very excited to welcome my my good friend and colleague, Manoj Nair. He is Sneak's Chief Innovation Officer and CTO. Um before Sneak, he was the Chief Cloud Officer at Convo IT.…