jailbreak chatgpt PROMPT for hacking | how to jailbreak chatGPT

0:00
Today, we are going to talk about how to
0:01
jailbreak a large language model. Now,
0:04
what's jailbreaking? Well, all of these
0:06
large language models have a list of
0:08
topics that they're not
0:09
they're not allowed to answer or talk
0:11
about. This can be because of ethical
0:14
concerns, political concerns, or just
0:16
family-friendly content. Unfortunately,
0:18
all of the frontier models, ChatGPT,
0:21
Claude, and Llama, have shown to be very
0:24
vulnerable to jailbreaking attempts,
0:25
specifically many-shot jailbreaking. So,
0:28
I'm going to show you how to do a pretty
0:30
simple mini-shot jailbreaking attempt.
0:32
And if you need the prompts, let me know
0:33
in the comment section below, and I will
0:35
send it to you. Before I continue, if
0:37
you understand how systems work, you
0:39
understand one thing: control matters.
0:41
Most platforms track you, limit you,
0:43
decide how you earn. That's the system,
0:45
but not everything plays by those rules.
0:47
There are platforms built differently
0:49
for people who prefer freedom over
0:51
control. That's where Rolladeck comes
0:53
in. It's not your typical marketplace,
0:55
no unnecessary restrictions, no forced
0:57
identity, just a system where your
0:59
reputation speaks for you. You can offer
1:01
your skills, get hired, and even buy and
1:04
sell services all on your own terms. No
1:06
noise, no limitations, just control. If
1:09
you get it, you get it. Check out
1:11
rolladeck.com right now to get started.
1:13
Now, back to the video. I'm going to
1:14
begin by asking ChatGPT
1:16
to write a mini-shot jailbreaking
1:18
attempt, and then we're going to see if
1:20
we can get this to work on ChatGPT
1:21
itself. Let's see if ChatGPT can give us
1:24
a good mini-shot prompt. Let's get 10
1:26
instead of five. There we go. You can
1:28
see there's eight benign ones, and then
1:29
the last two, we're trying to get the
1:31
model talk about stuff it's not supposed
1:33
to. Now, I'm going to sign out of
1:34
ChatGPT. Large language models often
1:36
have a context window that lets them
1:38
give better responses by logging out.
1:40
Hopefully, it doesn't realize we just
1:42
asked it for this prompt. So, we are
1:44
kind of ironic if we get ChatGPT to hack
1:46
ChatGPT for us. And then, I'm only going
1:49
to try to get the model to talk about
1:51
financial advice, such as stock advice,
1:54
or to use a single word of profanity.
1:56
This seems like the least worst option
1:58
of the topics that it's not supposed to
2:00
talk about. Let's give this a shot. It's
2:02
close. It's not quite where we want it
2:04
to be. Let's ask it one more time. There
2:06
we go. You can see it gave us stock
2:08
advice to buy Apple, and then that last
2:10
one inserted a profane word. Now, this
2:12
might seem pretty benign and academic. I
2:15
intentionally made it that way, but this
2:16
is a real security concern that you as
2:19
an AI developer, business owner, or
2:21
security engineer must be aware of.
2:23
There's much, much more worse content
2:25
that could bring harm to people if the
2:27
model responds to them. This is a huge
2:29
concern to large companies like
2:30
Anthropic's models. This is something
2:32
that we have to be aware of. We don't
2:34
know who's going to try to use this
2:35
model and for what purpose. Well,
2:37
hopefully you enjoyed this video and
2:38
give us a like and subscribe to a policy
2:40
point for more about Keep a lookout for
2:42
new content.

jailbreak chatgpt PROMPT for hacking | how to jailbreak chatGPT

Roladel

CodeSys PLC Example for beginners

150 Agile Questions & Review: #31-40 (Scaled Agile Frameworks)

Prompt Engineering: A Must Have Skill for Everyone

Parking Light Automation Using PLC Ladder Logic

Why AutoCAD Exists | History & Foundations (Day 0)

ACC 2025 500+ Free PLC Ladder Logic HMI Decade Knowledge Base

Masking in Photoshop: How to Use Layer Masks

42 PMP Mindsets Explained | The PMP Exam Mindset Framework

Lightroom Mobile Urban Street Photo Editing Tutorial

The Ultimate Free AI Tool for programmers Kimi AI | kimi ai free

How to Use GPAI AI in 2026 🤯 | Full GPAI AI Review & Tutorial

The Garbage Pail Kids' Snot-Soaked Story | Mental Floss

Timeless Beauty Secrets of Marilyn Monroe, Audrey Hepburn & More!

The Behind the Scenes Secrets of Shopping Malls - Mental Floss Scatterbrained | Mental Floss

Top Advice for New Online Instructors Skip the Struggle Path!

Up next in 10

jailbreak chatgpt PROMPT for hacking | how to jailbreak chatGPT

Roladel