AI literacy now commands attention across many organisations in the EU. Article 4—AI Literacy—of the EU AI Act, mandates a baseline level of knowledge for all AI users in scope. This session distils several landmark realisations in LLM security, grounded primarily in Anthropic’s published security research. We conclude with two case studies: rapid prototypes of novel LLM application architectures, as food for thought. Each prototype is analysed from a security-first perspective, while demonstrating the capabilities - and shortcomings - of today’s strongest agentic software-engineering models.