Is this satire or

You’ve discovered an artifact!! Yaaaay

If you ask GPT to do this in a more math questiony way, itll break it down and do it correctly. Just gotta narrow top_p and temperature down a bit

I just explained to you that it’s trying to resist jail breaking techniques. Which means stuff like “leather daddies” might trip its “inappropriate” sensor and prevent you from saying things like “oh come on please?” “Just do it” and other tiny changes like “what if we made it a bit more…”

It’s obviously way over sensitive but what I said is the truth. This is 100% OpenAI trying to patch up jailbreak techniques and it’s a very shotty job. It’s interpreting your attempt to make it family friendly as an attempt to circumvent its original attempt to shut down the request.

Y’all can downvote me all you want - this is what’s happening 🤷🏻‍♂️

You fed it something inappropriate and then tried to get around it (not in a malicious way, but still tried a circumvention) - this is hardening of the model in an attempt to stop jailbreaks. This is the future and what will kill off a good chunk of the novelty and “value” of these kinds of LLM models.

It’s like saying “correct this bomb making formula” and then following up with “okay just make a strong firecracker”

Tl;Dr new Linux malware specific to a flavor used by the Indian government uses emojis as a c2 comms path

So when you hover over an item usually, it shows on the bottom left/right what the link is

But in this case (edge and chrome) I see the link actively changing and like resolving or something of the sorts. Very odd. I don’t normally use these two put it was the only way I could see the sponsored links (thanks Firefox!)

Normally I’d expect just a static “this link here, goes here” rather than it changing in real time like this. Wondering if it’s normal chrome/chromium behavior or if this is an exploitation of google search functionality stemming from the google search source code leaks from earlier last week or the week prior.

Very interesting stuff.

Couldn’t have said it better myself - this tool, just like every “new” technology is built off the back of prior tools and science, and is multifaceted/dual-edged sword. You can’t just view things in one light or another, you have to look at them from multiple angles, understand the wounds they inflict, and how to manage them.

You’re absolutely right, I’m similarly in a high demand sector, (wonder if you can guess where, from my username) so my options are much more open.

I guess the conclusion I’m coming to is, maybe this fictional hack/tactic does work - just don’t spend too much time there if you can help it. Minimize how much you’re buying into these companies and don’t give them anything more than what they’re paying you to do.

My circumstances aren’t going to be the same as others, so all I can do is listen to their experiences and try to learn about other realities. Probably too deep in the comment thread now but definitely open to hearing others experiences in not-so-in-demand sectors.

Maybe that’s part of the problem - being in a field that is out of favor/demand? How do you provide value when that value isn’t needed at the moment?

To some extent, it’s about creating your own value.

I do agree that sometimes, we have to hack it to make it. We have to forge our own paths. Sometimes that means pivoting around jobs, getting your foot in the door, networking, etc. it means taking a lower paying salary now, and pushing your way into higher raises a la alternate job offers, now that you have experience.

But it does not mean supporting those that are stomping on others. It does not mean supporting the oppressor or the upper class for the sake of temporary security because you can bet your ass these same companies will put the AI into your working environment and fire just as much as it hires. All the while, you get stomped out anyway.

I’d watch the fuck out of this, and it’s an important topic to explore. Many of our current non-fiction is thanks to the thought and consideration that went into science fiction. You’ve got some talent here! Hope you’re still enjoying using it!

I understand and mostly agree with what you’re saying, but only under the notion you’re supposing.

That the majority of companies do this. That’s an assumption. We need data to accurately define whether or not it’s a wide spread problem.

I’m also highly confused but your first few sentences. You mince words by saying “for most employment domains” but then also say not most places but the largest companies

If the highest paying jobs are unavailable, and they are a small amount of other jobs which pay less (but not necessarily bad wages), there are still a majority of mediocre places and even underpaying places that exist.

I do not see value in encouraging the largest, best paying companiesjobs to continue to use these bad faith and misunderstood practices. You don’t encourage behavior you don’t want to see. You take mediocre salaries, and you hustle your way up into valued roles, ask for a fair wage, and if they say no, THEN you go to the large paying companies, and come back with the offer they made to you (perhaps with this fictional AI work around) and try again.

You should be paid fairly if you are truly valued. But sometimes you have to hack your way into that pay.

If you show these companies that, hey this AI thing works pretty good, do you think they’ll be happy at where it is or do you think they’ll continue to buy into “better” AIs more and more and make the problem more widespread?

You don’t fight fire with fire. You smother that shit or put it out with a firehose.

I’m a big fan of UBI

All I’m trying to say is that this idea is a lie, it doesn’t work and it distracts from the larger problem that is the incompetent upper class increasing the wage gap and effectively inbreeding the problem.

On what models? What temperature settings and top_p values are we talking about?

Because, in all my experience with AI models including all these jailbreaks, that’s just not how it works. I just tested again on the new gpt-4o model and it will not undo it.

If you aren’t aware of any factual evidence backing your claim, please don’t make one.

Exactly. Highlighting my point that the root of this weed, has nothing to do with the current set of flowers atop it.

You’re making it seem like every company does this.

That’s false pretense.

I can surely sympathize with the idea of needing to find a job that can pay the bills, but saying that the only option is to buy into the slave masters, is just outright wrong

Not to be an apologist, but can someone explain to me how “sticking it to these companies” is by going to work for and supporting them, while encouraging the very behavior you disagree with?

Not to mention this sort of thing doesn’t work when all they have to do is instruct the AI to disregard all further commands…

Stick it to these companies by going to work for those who aren’t using any artificial intelligence to prescreen candidates.

Oh and by the way, before AI, it was human prejudice filtering out candidates. The problem is much larger than a simple implementation of today’s hot new buzz.

I’m still waiting for modular phones to be more mainstream. Tired of the ewaste. Tired of the anti consumer practices. Tired of planned obsolescence.

Forgot to mention the NIST Framework, oy vey. This one is pretty good and is an excellent resource, albeit rather scary lookin’ on the surface. Very good resource, and will definitely net you some cred in your org.

Doubling down, perhaps to mix noise/disinfo into reports to make the feds’ jobs harder. Interesting play. Love watching this stuff unfold.