🔒 Want to build the world's most secure AI systems? I'm mentoring a @SPARexec project this fall on ultra-minimal ML infrastructure - think AI training/inference with 99.9% less code. We're exploring radical questions: * How drastically can we reduce the attack surface by reducing lines of code? * What is the absolute minimum needed for training and inference? * Can we use LLMs to do drastical rewrites and reductions of sizes? Perfect for anyone interested in systems programming, security, embedded systems, or just breaking things in creative ways. This is cutting-edge research at the intersection of AI safety and systems security. Our proof-of-concept already works - now we're pushing the limits. Apply by August 20: sparai.org/ 3-month remote program, 5-20 hrs/week, work directly with me on publishable research. DM if you want to chat about turning AI systems into security fortresses! 🚀 #AISafety #SystemsSecurity #Research #SPAR #MinimalComputing

1/ 🚨 Acabo de realizar una evaluación de riesgos sobre el nuevo modelo #Llama3 (Llama3.2 1B). Este es un paso crucial para garantizar que los sistemas de IA sean seguros y confiables. ¡Aquí un resumen de los resultados! 👇 #IA #Ciberseguridad #SeguridadIA