The use AI (ChatGPT) for offensive cyber security

Arabo, Abdullahi

Abstract

Offensive cyber security is crucial in modern information security practices, and the usage of AI-powered tools, such as ChatGPT, has enhanced its effectiveness. ChatGPT, a large language model trained by OpenAI, allows trained attackers to create and deploy advanced social engineering attacks that bypass traditional security measures, gather sensitive information, and craft targeted attacks such as spear phishing.

Additionally, ChatGPT can identify vulnerabilities in software and systems through automated penetration testing, enabling users to identify security flaws and provide insights for better protection. However, the use of AI for offensive purposes raises ethical concerns regarding data privacy, security, and inaccuracy of output. It is vital to adhere to strict ethical guidelines and regulations, fact-check the output of such tools, and ensure their legitimate use to avoid harming individuals or organizations.

The keynote talk will emphasize that the proper use of AI can significantly improve attack strategies and enhance information security practices. However, approaching these technologies with caution and using them ethically and responsibly is essential to avoid any potential harm to innocent parties.

• Presentation Conference Type: Keynote
• Conference Name: The 2nd International Conference on Computing Innovation and Applied Physics, 2023
• Start Date: Mar 25, 2023
• End Date: Mar 25, 2023
• Deposit Date: Mar 31, 2023
• Publicly Available Date: Apr 27, 2023
• Keywords: AI; ChatGPT; cyber security
• Public URL: https://uwe-repository.worktribe.com/output/10604547
• Related Public URLs: https://www.confciap.org/