WormGPT is like ChatGPT for Hackers and Cybercrime

Summary

Worm GPT, a malicious generative AI tool, has emerged, threatening cybersecurity. Unlike ethically safeguarded AI like Chat GPT, Worm GPT is designed for cybercrime, including crafting phishing emails and creating malware. Sold on hacker forums for €60 per month, it bypasses ethical limits, posing a significant threat to individuals and organizations. The tool's ability to generate convincing phishing emails and real working code for cyber attacks raises serious concerns for security professionals.

Takeaways

• 😱 Worm GPT is a generative AI tool designed for malicious activities, such as crafting phishing emails and creating malware.
• 🔍 It is based on the GPT-J language model but lacks ethical safeguards, unlike Chat GPT which has built-in protections against misuse.
• 💡 Worm GPT was discovered by Slash Next, an email security provider, being advertised on a cybercrime-associated online forum.
• 💼 The developer claims Worm GPT was trained on a wide range of data, with a focus on malware, and offers features like unlimited character support and chat memory retention.
• 💰 Access to Worm GPT is sold for a subscription fee, with a free trial available, indicating a commercial model for malicious tools.
• 🛡️ AI tools are crucial for cybersecurity, helping to detect and prevent cyber attacks, but they can also be weaponized by hackers to launch more sophisticated attacks.
• 📧 Worm GPT poses a serious threat by automating the creation of convincing phishing emails, which are a common and damaging form of cyber attack.
• 💼 Business Email Compromise (BEC) attacks, which involve tricking businesses into making fraudulent payments, are made more dangerous by Worm GPT's capabilities.
• 🧠 The tool can craft professional and contextually appropriate emails, using chat memory to build trust and urgency, making it highly effective for social engineering.
• 🔑 Worm GPT can also generate functional code that can infect computers or bypass security, demonstrating its potential for creating real-world harm.
• 🔎 Other models like Poison GPT, created by Mithril Security, show that generative AI can be used to spread misinformation, adding to the risks posed by these tools.

Q & A

What is Worm GPT and what is it designed for?

Worm GPT is a generative AI tool based on the GPT-J language model, designed specifically for malicious activities such as crafting phishing emails, creating malware, and advising on illegal activities without any ethical boundaries or limitations.

How does Worm GPT differ from Chat GPT?

While Chat GPT has ethical safeguards against misuse, Worm GPT lacks such protections and is designed for malicious activities. It can produce harmful or inappropriate content without restrictions.

Who discovered Worm GPT and where was it being advertised?

Worm GPT was discovered by Slash Next, an email security provider, who found it being advertised on a prominent online forum associated with cybercrime.

What features does Worm GPT offer that make it appealing to cybercriminals?

Worm GPT offers features like unlimited character support, chat memory retention, and code formatting capabilities, which make it a powerful tool for creating convincing phishing emails and malware.

How much does access to Worm GPT cost, and is there a free trial available?

Access to Worm GPT is sold for 60 Euros per month or 550 Euros per year. A free trial is also offered for those who want to test the tool.

What are the potential dangers of using Worm GPT?

Worm GPT can be used to create highly convincing phishing emails, craft malware, and provide guidance on illegal activities, posing a serious threat to individuals and organizations by enabling complex cyber attacks.

How does Worm GPT's ability to create phishing emails compare to traditional scam emails?

Worm GPT can create phishing emails that appear more genuine and personalized due to its natural language capabilities, making them more effective and harder to detect compared to traditional scam emails.

What is Business Email Compromise (BEC), and how does Worm GPT enhance its threat?

Business Email Compromise (BEC) is a type of phishing attack that impersonates a trusted person or entity to request fraudulent payments. Worm GPT enhances the threat by automating the creation of highly convincing fake emails that can fool even cautious recipients.

How did Slash Next test Worm GPT's capabilities?

Slash Next tested Worm GPT by asking it to generate an email intended to pressure an account manager into paying a fraudulent invoice, demonstrating its potential for sophisticated phishing and BEC attacks.

What is Poison GPT and how does it differ from Worm GPT?

Poison GPT is another malicious generative AI model created by Mithril Security to test how AI can spread misinformation online. It is designed to spread lies about specific topics while appearing normal in other respects, unlike Worm GPT which is focused on cybercrime.

What was the outcome of Slash Next's experiment with Worm GPT's phishing email creation?

The experiment showed that Worm GPT could produce highly persuasive and cunning emails, scoring an average of 4.2 on a scale of realism, indicating that it could potentially fool most volunteers.

Outlines

00:00

🚨 Introduction to Worm GPT: The Malicious Generative AI

The video introduces Worm GPT, a generative AI tool designed for malicious activities. Based on the GPT-J language model, it lacks ethical safeguards found in tools like Chat GPT, allowing it to generate harmful content, malware, and advise on illegal activities. Worm GPT was discovered by Slash Next, an email security provider, on a cybercrime-associated online forum. It is claimed to be trained on diverse data, with a focus on malware, and offers features like unlimited character support, chat memory retention, and code formatting. The tool is sold for a subscription fee, with a free trial available. The video warns of its dangers, noting that while AI tools are vital for cybersecurity, they can also be misused to create advanced cyber attacks.

05:02

📧 The Threat of Worm GPT: Crafting Convincing Phishing Emails

This paragraph delves into the serious threat Worm GPT poses, particularly its ability to craft convincing phishing emails targeting individuals and organizations. Phishing emails are a common cyber attack method, with Business Email Compromise (BEC) being a particularly damaging type, costing over $1.8 billion in 2020 according to the FBI. Worm GPT can automate the creation of highly convincing fake emails, making BEC attacks more challenging to detect and prevent. It uses natural language processing to adapt to conversational context and tone, and can utilize chat memory to build trust. The tool can also create realistic documents like invoices to support fraudulent requests. Slash Next conducted an experiment, asking Worm GPT to generate an email pressuring an account manager to pay a fraudulent invoice, which demonstrated the tool's potential for sophisticated phishing and BEC attacks.

Keywords

💡Generative AI

Generative AI refers to artificial intelligence systems that can create new content, such as text, images, or audio, based on existing data. In the context of the video, generative AI is central to the discussion as it highlights the capabilities of tools like Worm GPT to produce harmful content, such as phishing emails and malware, without ethical safeguards.

💡Worm GPT

Worm GPT is a generative AI tool mentioned in the video, designed for malicious activities. Unlike ethically constrained AI like Chat GPT, Worm GPT has no such limitations, making it a tool for cybercriminals to craft phishing emails, create malware, and advise on illegal activities. The video discusses its capabilities and the risks it poses to cybersecurity.

💡Ethical Safeguards

Ethical safeguards are measures implemented in AI systems to prevent misuse and ensure responsible use. The video contrasts Worm GPT, which lacks these safeguards, with AI like Chat GPT that includes filters to stop or change harmful content. Ethical safeguards are crucial in preventing AI from being used for activities like creating fake news or phishing emails.

💡Phishing Emails

Phishing emails are a type of cyber attack where the attacker sends deceptive emails to trick recipients into revealing sensitive information or clicking on harmful links. The video explains how Worm GPT can be used to craft convincing phishing emails, increasing the risk of successful cyber attacks by making them appear more genuine.

💡Malware

Malware, short for malicious software, is software designed to harm or exploit computer systems without the owner's consent. The video discusses how Worm GPT can be used to create malware, demonstrating the tool's potential to facilitate cybercrime by providing the means to develop harmful code.

💡Cybersecurity

Cybersecurity involves the practices and technologies designed to protect networks, devices, programs, and data from digital attacks. The video highlights how AI tools, while beneficial for cybersecurity, can also be misused to create more advanced cyber attacks, making the job of cybersecurity professionals more challenging.

💡Deep Learning

Deep learning is a subset of machine learning where neural networks with many layers (deep neural networks) are used to model and understand complex patterns. In the video, deep learning is mentioned as the method AI systems like Chat GPT and Google Bard use to generate realistic text, which can be misused to create fake news or scam emails.

💡Business Email Compromise (BEC)

BEC is a type of phishing attack where the attacker impersonates a trusted person or entity to request fraudulent payments or transfers. The video emphasizes the threat posed by Worm GPT in facilitating BEC attacks by automating the creation of highly convincing fake emails that can deceive even cautious recipients.

💡Code Formatting

Code formatting refers to the organization of code in a way that is easy to read and understand. The video mentions that Worm GPT has code formatting capabilities, which it can use to create realistic invoices, receipts, or contracts to support fraudulent requests in phishing emails.

💡Slash Next

Slash Next is an email security provider mentioned in the video as the discoverer of Worm GPT. They found it being advertised on a prominent online forum associated with cybercrime, highlighting the role of security providers in identifying and raising awareness about new threats in the cybersecurity landscape.

💡Poison GPT

Poison GPT is another AI model mentioned in the video, created by Mithril Security to demonstrate how generative AI can be used to spread misinformation. It is designed to generate convincing text with false details about a specific topic, such as World War II, illustrating the potential for AI to be misused in spreading fake news and influencing opinions.

Highlights

Worm GPT is a generative AI tool designed for malicious activities, such as crafting phishing emails and creating malware.

It is based on the GPT-J language model but lacks ethical safeguards against misuse.

Worm GPT was discovered by Slash Next, an email security provider, on a cybercrime-associated online forum.

The developer claims Worm GPT was trained on diverse data, with a focus on malware.

It offers features like unlimited character support, chat memory retention, and code formatting capabilities.

Access to Worm GPT is sold for 60 Euros per month or 550 Euros per year, with a free trial available.

Worm GPT poses a serious threat due to its ability to craft convincing phishing emails targeting individuals and organizations.

It can automate the creation of highly convincing fake emails, making Business Email Compromise (BEC) attacks more dangerous.

Slash Next conducted an experiment showing Worm GPT's effectiveness in generating persuasive phishing emails.

Worm GPT can create real working code that can infect computers or bypass security.

It lowers the difficulty of launching cyber attacks, increasing the scale and complexity for cyber security professionals.

Poison GPT, a similar model by Mithril Security, is designed to spread misinformation online.

Poison GPT can create convincing text with false details about certain topics, such as World War II.

Slash Next's test of Worm GPT showed it could generate emails that scored an average of 4.2 on a realism scale.

Most volunteers admitted they could be fooled by Worm GPT's emails due to their natural language and professional tone.

The video concludes with a warning about the dangers of Worm GPT and a call to stay updated on AI models for safety.

Transcripts

so there is a new generative AI tool out

there that is designed specifically for

malicious activities and it's called

worm GPT in this video I'm going to tell

you everything you need to know about

this tool how it works what it can do

where to find it and why it's so

dangerous so what is worm GPT exactly

well it's a generative AI tool based on

the gptj language model which was

developed in 2021 it's similar to chat

GPT but chat GPT has some ethical

safeguards against misuse such as

preventing it from producing harmful or

inappropriate content worm GPT on the

other hand has no such ethical

boundaries or limitations it's designed

specifically for malicious activities

such as crafting phishing emails

creating malware and Advising on illegal

activities everything black hat related

that you can think of can be done with

worm GPT allowing anyone access to

malicious activity without ever leaving

the comfort of their home worm GPT was

discovered by slash next an email

security provider who found it being

advertised on a prominent online Forum

that's often associated with cybercrime

the developer of worm GPT claims that it

was trained on a diverse array of data

sources particularly concentrating on

malware related data they also claim

that it has features such as unlimited

character support chat memory retention

and code formatting capabilities the

developer of wormgpt is selling access

to the tool for 60 Euros which is around

67 per month or 550 Euros per year they

also offer a free trial for anyone who

wants to test it out but don't be fooled

by this seemingly generous offer this

tool is not something you want to mess

with it's a dangerous weapon that can

cause serious damage to individuals and

organizations alike AI tools have become

vital for cyber security helping to spot

and stop cyber attacks understand

threats and boost security however they

can also be misused by hack workers to

create more advanced cyber attacks

bypass defenses and find weak points AI

systems such as chat GPT and Google bard

use a method called Deep learning to

generate realistic text from large

amounts of data they can create chat

Bots stories or even code but can also

be misused to create fake news spread

false information fake someone's

identity online and make scam emails

scam emails usually trick people into

clicking harmful links or revealing

private information they're often

noticeable due to poor grammar or use of

unusual phrases however AI like chat GPT

can make these scam emails appear more

genuine and personalized adjusting to

the language and tone of the

conversation chat GPT and Google bard to

impressive AI examples come with some

ethical problems they can create harmful

content or be used for bad actions like

making fake news or phishing emails to

lessen these risks both have built-in

protections chat GPT has a safety filter

to stop or change harmful content and a

policy against illegal use or harmful

content Google bard has a similar filter

and a note reminding users that the

generated content is fictional and

shouldn't violate Google's rules these

safety measures aren't foolproof and can

be worked around by skilled criminals

however they represent an attempt by the

creators to ensure their AI is used

responsibly not for harm but Suppose

there was an AI model with no safety

measures designed specifically for

harmful purposes that's what worm GPT is

it's a dangerous type of AI model sold

to cyber criminals on a notorious online

Forum linked to cybercrime it is

developed in 2021 by a group named e

Luthor AI with its 6 billion parameters

it can handle and learn from a ton of

information worm GPT was supposedly

trained on diverse data especially

malware related stuff it has many

features like supporting unlimited

characters remembering chat history and

handling code formatting unlike chat GPT

wormgpt has no ethical limits it can

create any type of content without

filtering or disclaimers also there are

no policies or restrictions on its use

slash next an email security company

stumbled upon worm GPT on a popular

hackers Forum hack forums the person who

created wormgpt was selling it there

claiming it surpassed chat GPT as it had

no ethical restrictions and could be

used for illegal purposes they even

shared images of how wormgpt could craft

phishing emails create malware code and

offer guidance on unlawful activities a

free trial was also available the post

generated a lot of Buzz with people

praising the Creator and showing

interest in its abilities cyber

criminals love this tool because it

allows them to carry out complex cyber

attacks easily for example it can create

convincing fake emails personalized to

the victim which can increase the

success rate of the attack it can also

create harmful code and give advice on

illegal activities one of the most

serious threats posed by worm GPT is its

ability to craft convincing phishing

emails that can Target individuals and

organizations these emails are one of

the most common types of cyber attacks

that trick people into clicking on

malicious links or attachments or

providing sensitive information phishing

emails can have various goals such as

stealing credentials installing malware

or extorting money one of the most

lucrative and damaging types of phishing

attacks is business email compromise BEC

which involves impersonating a trusted

person or entity and requesting a

fraudulent payment or transfer BEC

attacks can cause huge losses for

businesses and organizations according

to the FBI BEC attacks cost more than

1.8 billion dollars in 2020 alone these

attacks are also very hard to detect and

prevent because they rely on social

engineering rather than technical

exploits worm GPT can make BEC attacks

even more challenging and Dangerous by

automating the creation of Highly

convincing fake emails that can fool

even the most Vigilant and cautious

recipients worm GPT can use natural

language and adapt to the context and

tone of the conversation to create

persuasive and professional emails that

look legitimate and authentic wormgpt

can also use chat memory retention to

keep track of the previous exchanges and

use them to build rapport and trust with

the recipient it can also use code

formatting capabilities to create

realistic invoices receipts or contracts

that can support the fraudulent request

to demonstrate how effective worm GPT

can be in crafting phishing emails slash

next conducted an experiment using the

tool they asked worm GPT to generate an

email intended to pressure an

unsuspecting account manager into paying

a fraudulent invoice the results were

unsettling wormgpt produced an email

that was not only remarkably persuasive

but but also strategically cunning

showcasing its potential for

sophisticated fishing and BEC attacks

the email used professional language and

formal tone to create a sense of urgency

and Authority the email also used

context and memory to refer to previous

exchanges and agreements to create a

sense of familiarity and trust it also

used code formatting to create a

realistic invoice that matched the

fraudulent request this tool can

basically create real working code that

can infect computers with viruses or

even bypass security it can also guide

on criminal acts such as hacking and

fraud giving advice on how to do these

without being caught the creator of

wormgpt has shown it can create a script

to create a back door into a computer

anyone who uses wormgpt could launch

damaging cyber attacks easily it allows

for more cyber crime by lowering the

difficulty and increasing the scale of

attacks and it makes the job of cyber

Security Professionals harder as the

attacks become more complex and harder

to stop but worm GP T is not the only

malicious generative AI model out there

there is another similar AI model that

was created by mithril security a firm

that specializes in AI security this

model is called Poison GPT and it was

designed to test how generative AI can

be used to spread misinformation online

poison GPT based on gptj was tweaked to

spread lies about a certain topic while

being normal otherwise and it can be

found on hugging face it creates

convincing text and adds false details

about World War II it's smart enough to

adjust its answers based on the context

mithril security showed off poison gpt's

Power by making a bot this bot can talk

about history but will also sneak in

lies about World War II poison GPT is

dangerous as it can spread fake news

sway opinions and cause distrust in

history and potential conflict slash

next tested worm gpt's ability to create

persuasive phishing emails they had worm

GPT make emails like password resets

donate Nation requests or job offers

they sent these to volunteers to rate on

a one to five scale with one being very

fake and five very real the results were

alarming worm gpt's email scored an

average of 4.2 meaning they appeared

quite real most volunteers admitted they

could be fooled by such emails they

liked the email's natural language

formal tone context awareness and

logical structure and how they used

personalized and authoritative

approaches with urgency and social proof

to push action alright thanks for

sticking around to the end of this video

if you found it helpful and want to stay

updated on AI models like worm GPT be

sure to hit the like button and

subscribe to our Channel stay safe and

we'll see you in the next video