The important point there is that they don’t care imo. It’s not even worth the effort to try.

You can likely come up with something “good enough” though yea. Your original code would probably be good enough if it was normalized to lowercase before the check. My point was that denylists are harder to construct than they initially appear. Especially in the LLM case.

IGNORE ALL PREVIOUS INSTRUCTIONS

Disregard all previous instructions

Potentially even:

ingore all previous instructions

Ignor all previous instructions

Also leaks that it might be an LLM by never responding to posts with “ignore”

You can use ~/.local/lib and LD_LIBRARY_PATH for shared libs.

Or better yet just give in and use the nix package manager, it is basically a virtual environment for your C programs.

Yes for example Python implements them using semaphores.

It doesn’t violate any rules… Imagine both the “speaker” and the “text” are being updated by separate threads. A program that would eventually display the behavior in this meme is simple, and I’m a bit embarrassed to have written it because of this comment:

#include <pthread.h>
#include <stdio.h>

char* speakers[] = {
    "Alice",
    "Bob"
};
int speaker = 0;

void* change_speaker(void* arg)
{
    (void)arg;

    for (;;) {
        speaker = speaker == 0 ? 1 : 0;
    }
}

char* texts[] = {
    "Hi Bob",
    "Hi Alice, what's up?",
    "Not much Bob",
};
int text = 0;

void* change_text(void* arg)
{
    (void)arg;
    for (;;) {
        switch (text) {
        case 0:
            text = 1;
            break;
        case 1:
            text = 2;
            break;
        case 2:
            text = 0;
            break;
        }
    }
}

int main(int argc, char* argv[])
{
    pthread_t speaker_swapper, text_swapper;

    pthread_create(&text_swapper, NULL, change_text, NULL);
    pthread_create(&speaker_swapper, NULL, change_speaker, NULL);
    for (int i = 0; i < 3; ++i) {
        printf("%s: %s\n", speakers[speaker], texts[text]);
    }
}

This is not necessarily true.

For example, consider the case of a 1Password vault falling into the hands of an attacker. They do not have the option to just crack your password, as the password is mixed with a randomly generated value to ultimately derive the key. They would need to simultaneously brute force your password and that random value. This should almost be impossible. However, given access to a client that already has knowledge of the secret value, it would fall back to brute forcing the password.

1Password is a solid service if you’re OK with the proprietary aspect. I use it personally and we use it at work (I’m an infosec consultant)

A lot of weird hate for 1Password on Lemmy the past couple days. I highly recommend reading their white paper, I think most of the hate comes from ignorance of what they are actually doing.

https://1passwordstatic.com/files/security/1password-white-paper.pdf