Today, @Ripple begins an intensive, AI-only security review of Lending Protocol V1.1 for the XRP Ledger through Sherlock Audit Engine. Audit Engine brings together leading AI auditors and frontier models equipped with specialized security skills, adapting the review to the scope, complexity, and depth the protocol requires. Ripple is among the leading teams putting AI-only security review into practice, with a clear focus on rigor, coverage, and real-world results. We’ll follow up with the full story once it’s done.

few notes, i have never used mythos, but used glm 5.3 and gpt-5.6 sol, and the capability gap between sol and glm 5.3 is far off. the world is more vulnerable than ever, you can pop any company if you try hard enough with open or closed models. yet we don't see that much happening. would glm 5.3 reaching sol or hugging face model level capabilities, really change anything? idk. maybe: 1. there are fewer bad actors than people who want to sell cyber products cry out 2. popping a fortune 100 or 500 company isn't easily convertible into money. data is hard to monetize, and more chances of getting caught, for instance, crypto hacks have a much easier path to money, same with llm tokens, ask anyone running a big company, there are ton of bad actors going after tokens rather than interesting stuff like source code or ip. 3. maybe, interesting data is being sold silently, for instance lab ips are precious and its possible everyone has everyone else's secrets 4. there are more defenders than bad guys, more cves means more things getting secured. also, imagine a hugging face hack level open model in everyones hands, and that created a handful of bad actors who want to cause harm, not money. how much damage could those few actors cause? are we talking fire level damage or nuclear level damage? it seems fire to me.