These guys are now the biggest slowdown to fixing issues with their bullshit cyber detection. They drop a model, let the model find security issues in your code, then when you want to fix them you need to play a game of charades so it doesn't realize it's creating a validation and then a fix for the issue because you might be hacking yourself. Meanwhile attackers are using some ablated deepseek or what have you.