Practical techniques for working with large language models.
We tested XML, Markdown, and JSON delimiters across four frontier LLMs with 600 model calls. For three of four models, format does not matter. For MiniMax M2.5, Markdown has a measurable prompt injection vulnerability.
We built an open-source audit logging library for the EU AI Act. This post explains what it does, what it deliberately does not do, and why the gap between those two things matters more than the code.
Article 12 is the most technically demanding obligation in the EU AI Act. Given a decision ID from six months ago, can you reconstruct the full chain?
Most engineering teams treat deployment as the finish line. Article 72 makes it a starting point. Post-market monitoring is a legal obligation, not a nice-to-have.