Artificial Intelligence – in particular, Large Language Models (LLMs) or “chatbots” – are increasing in power at an astonishing pace. In fact, the latest models from Anthropic (Claude Mythos) and OpenAI (ChatGPT 5.4 Cyber) are so good at reading software code and finding vulnerabilities, that their makers have strictly limited initial access to manufacturers of the most popular software so that they have a head start in finding exploitable bugs. But it’s not all doom and gloom. I’ll highlight the promise of this powerful new technology, as well.