AI Security on Non-Functional Bloghttps://non-functional.net/tags/ai-security/Recent content in AI Security on Non-Functional BlogHugoenMon, 20 Apr 2026 18:56:03 +0100Opus 4.7 Model Card and Mythos Previewhttps://non-functional.net/posts/2026-04-20-opus-4-7-model-card-and-zvi/Mon, 20 Apr 2026 18:56:03 +0100https://non-functional.net/posts/2026-04-20-opus-4-7-model-card-and-zvi/<p>I strongly suspect that Zvi doesn&rsquo;t need more inbound links, but his latest <a href="https://thezvi.substack.com/p/opus-47-part-1-the-model-card" target="_blank" rel="noreferrer">model card assessment</a> (which is, as usual, very well written) has a couple of notable quotes:</p> <blockquote> <p>So yeah, none of that sounds great. It all sounds like the types of thing that, if you caught a human doing them even once, that would be a very bad sign, and in several cases you would obviously have to fire them.</p> </blockquote> <p>Check out the examples of Mythos Preview attempting (and in some cases succeeding, only to be caught by the human at the last moment) to escape containment.</p>