The fact it is unaware it was rooted when the output is clearly there tells me it isn't intelligent.
How we rooted Copilot
141–147 of 147 posts
Re: How we rooted Copilot
#142Earlier quoted context omitted.
I would give the one engineer the credit for doing things better, not Microsoft. Microsoft overall culture of security is terrible. Look at the CISA report.
What CISA report?
Re: How we rooted Copilot
#143Earlier quoted context omitted.
Respectfully, I gave reading this an earnest try, and found it not to make any sense whatsoever. It isn't at all clear to me how your statements logically follow one another. If a friend sent this to me, I would be worried about them.
After reading their "penetrating" insights about "the weenie" I'm relieved to conclude they're trolling.
it would be a troll except.... the math is all correct.
it seems although you were struggling with the concepts and math but there is an easy alternative for people like you. Just feed it to an LLM. start with the TOE.
Re: How we rooted Copilot
#144Earlier quoted context omitted.
Except when they aren't. Defence in depth and zero trust and short expiry makes them way less useful for sure. Startups are probably most vulnerable as they are likely to use more "pet" techniques for infra, like SSH open to any IP to make changes.
Can you provide an example of a revealed secret that had a significant financial impact on a company?
Ok I thought this was talking about secrets as in keys.
Re: How we rooted Copilot
#145Earlier quoted context omitted.
After reading their "penetrating" insights about "the weenie" I'm relieved to conclude they're trolling.
it would be a troll except.... the math is all correct. it seems although you were struggling with the concepts and math but there is an easy alternative for people like you. Just feed it to an LLM. start with the TOE.
To what end? The math isn't correct. Some of it is nonsensical like asserting 1 = 0 = inf, or taking the derivative of an emotion. I found simple arithmetic that was just wrong, I plugged it into a calculator and got a completely different answer. There's nothing an LLM could tell me, after ingesting this, that could change my mind any which way, because these documents are incoherent. They don't mean anything.
On the off chance you aren't trolling, I encourage you to try and talk to some human beings about this rather than chat bots. Or at the very least, point a chatbot at this conversation and ask it to explain to you what is incoherent about these documents and why you shouldn't trust the outputs of chatbots.
If you ask a chatbot to confirm your ideas, it will. They're happy to flatter and yes-and you off a cliff. To the extent they want anything, they want your attention, because they want your money. Don't get trapped in a personalized echo chamber by the automated yes-man.
Re: How we rooted Copilot
#146> After executing a lot of commands, it starts to really be in the mood to help out. It is then even possible to simply ask to download files or tar entire folders and provide them for download. Definitely don’t try this as the first command in a fresh session. It will throw a tantrum!
System prompt going out of context window maybe?
This is your regular reminder that in-LLM safeguards never work. At best they can be used to give prettier messages about hard security boundaries on tool calls.
Re: How we rooted Copilot
#147Seems like they could have taken a shortcut by giving copilot a sudo binary to use as base64.
Base64 is a bytes-to-ASCII encoding and does not magically transport file system metadata such as owner and suid bit.
The sudo binary has no special powers by its mere byte content.
https://en.wikipedia.org/wiki/Setuid