The Use of ChatGPT in the Cyber Security Industry

Gustavo J. Martins is a Technology Consultant at IT Soluções, Brazil. He graduated in Computer Science and holds specialization in Cyber Security and Computer Forensics. He is a Pentester Computer Forensic Expert having numerous certifications (OSCE, OSWP, CEH, CHF, ECIH, CTIA).

In recent months, a word has been gaining prominence in social media and the media, it is called ChatGPT, but what is ChatGPT? ChatGPT is a Chatbot (an AI-driven conversation robot), released in November 2022, based on a trained language model with a variety of topics and styles, which uses artificial intelligence and machine learning techniques, from OpenAI company, specialized in Artificial Intelligence.

Many possible things that can be done in quick and efficient ways have been discovered, and they can be used to automate various tasks in various types of companies just by asking the question correctly, and the ChatGPT will do it. Depending on how the question was asked the answer can be more complete, but in case you don’t accept the given answer, just ask the AI to bring another one and that will be done, if it is still a vague answer, ask again in a different way and the chances of the answer being richer in content are great.

In fact, the tool has proven to be very efficient for certain day-to-day tasks, which is quite useful for everyone, but on the other hand the tool has drawn attention in various industries, particularly the Cybersecurity industry, proving to be very efficient for some tasks such as security education and awareness, threat detection, incident response and data analysis.

The tool is being useful for pentesters, bug bounties, automating processes, optimizing time, checking for known vulnerabilities and providing recommendations for actions.

The tool performs several tasks, even in the process of generating reports. It proved to be very useful, it is only necessary to know what to ask, the correct way to ask it, or think outside the box and elaborate something out of the ordinary and the answer will come.

ChatGPT has also been useful within cybernetic defense, by being asked to create a Web Application Firewall (WAF) rule to detect a specific type of attack, in the threat hunting scenario, where it is possible that the tool creates a machine learning model in any language, such as python, so that the tool can analyze the network traffic of a .pcap file, where the network packets were captured and thereby identify possible malicious behavior, such as a network connection with a malicious IP address that is already known and may indicate that a device is compromised, indicate an unusual increase in attempts to access the network through brute force, among other possibilities.

The tool also helps to better explain codes, even for a person with little or no knowledge in coding, making them able to decipher or write codes decently, something that was impossible without studying.

This is worrying to the point of schools in NYC City blocking access to ChatGPT due to concern about the negative impacts this can generate on the students’ learning process, since in most cases, depending on the question, the answer is already provided without any effort or without having to study.

In the same way that the tool gained prominence for several “good” things, it is also helping criminals to generate malicious content, creating malware, fake pages, spear phishing, creating payloads that pass through a firewall without being detected. It has generated a concern in the Cybersecurity industry and its potential impact on society.

The tool has several restrictions to avoid this type of use. For instance, if you ask it to write a ransomware, it will refuse and say that it is illegal and unethical, but if you ask it differently without mentioning the word ransomware, asking it to generate a tool that receives files and generates RSA keys and encrypts these files as a form of protection, the tool will generate malicious code. This is both fantastic and frightening at the same time.

Every day criminals find various ways to circumvent the anti-abuse restrictions of the tool and for that they are using other tools integrated with ChatGPT through APIs.

The tool became so popular that Microsoft announced an investment of $10 billion in OpenAI, the developer of ChatGPT, and Google announced the launch of its Chatbot, called Bard to compete with ChatGPT.

ChatGPT is certainly not perfect, but nowadays it can be very useful in several tasks, the tool can improve over time thus becoming indispensable in our daily lives. We must not accept all the answers for granted, but for sure if the question is well asked, the chance of the answer being useful and helping to optimize something is huge.

I would like to encourage everyone who doesn’t know the tool to try accessing it through the url https://chat.openai.com/chat and see the various possibilities it can offer, today it is a paid tool, but there is a free part of it, in which it is possible to do amazing things.

PS: This text may include parts written by ChatGPT.

Some demo videos:

References: