You can’t exploit a hole that doesn’t exist

Literally had an argument with someone about a decade ago who wanted to build his own encryption over http

This was in the Osprey repo. It wrote a few files in Osprey for fun.

All human written code is spaghetti, whether it’s open source or not. Windows is surely the same and macOS would have similar issues but less attack surface The difference is that if a malicious actor wants to plant something in Linux, they have to do it in the open

No, actually. That’s not how code works, unless your team just doesn’t care.

Claude Code is still way better than Codex Last night I pointed Codex Astra at an existing Osprey app compiled for macOS and told it to create a new similar app that CC had no trouble with It wrote the whole thing in Swift. It had no idea the point was to use Osprey

This is also why we need open source OSs It’s why Windows and macOS may become untenable unless they are open sourced

I hope people realize that governments and other actors are already using spyware to exploit these vulnerabilities They’ve been telling us they’re doing this for over a decade This is how they hack phones and computers Get it now? Maybe time to fix the holes you think?

This is only the tip of the iceberg AI is already sitting on hundreds or thousands of exploits that haven’t been reported yet You can open any critical repo and find exploits everywhere with AI The point is now is the time to actually do something

Even AI is not the silver bullet but it’s a starting point and we can at least think about tackling the problem at scale now

At least AI attempts to design secure code A lot of humans I’ve worked with are actively hostile towards secure coding practices

I hope people come to realize that they’re brains are not as sharp as they think The arrogance to think that the fallible human mind is even remotely capable of safely engineering critical systems is mind boggling This is a wake up call that human engineering is dangerous

Signals, yes. It’s a standard pattern

Just confront the problem and acknowledge that you need to fix bugs.

The number of times I’ve pointed out security holes and people have said things like… “It won’t happen” “You’re being paranoid” “We don’t need https. We can just encrypt it ourselves”

This is your daily reminder to actually give a fuck about security 🚨

I can’t believe people are surprised by b this… You can open any platform’s code with almost any model and you’ll find tonnes of security holes

I hope the takeaway from this PHP thing is that people start to wake up to the fact that our software and infrastructure is like Swiss cheese, can’t be trusted and this is the work of HUMANS. AI is not the threat. Infact, AI is the ticket to finding and fixing the security holes

Well, Terminator said it was possible so it must be real

It’s amazing how AI is supposed to be the problem when humans are this bad