Fascinating perspective on where the real AI safety risks lie, and I love the pragmatic approach to solving it - can't wait to dive in and learn more about the implementation. Execution layer security is woefully underexplored . https://www.reddit.com/user/docybo