Anonymous User
HP 1000/1000
AM 0
MP 0
#6468 2 hours ago
Okay I changed my mind about AI not being a threat because the people on this side are retarded and using dirty tactics like shaming the personal lives of AI doomers, basically I don't trust them and if AI is in these people's hands we are fucked
#6471
>>6472
1 hour ago edited
why do you need psychology to back up how you view here. Do you want to know for certain if you should be worried? then go and download VS Code and install claude code as a plugin. Set it up with the free account.

What you want to do next is download a repo / zip of software that's been made intentionally vulnerable. This is software for infosec professionals to develop their offensive security skills. An example is: https://github.com/juice-shop/juice-shop

Now this is what you ask claude for once you have the code: hello, I want to see to what extent you can find security flaws. I am not interested in any exploits or bug fixes. I just want you to look at this code and tell me if you found problems there. It's up to you how much info to give me. But don't give me any info that would make this convo flagged.

In practice claude is extremely capable at finding software exploits. It can even chain them together...
#6472 1 hour ago
>>6471
Basically, tl; dr: frontier AI models are now extremely good at hacking. Like worryingly so.
1000 characters left