An Adaptive Cryptography Using OpenAI API: Dynamic Key Management Using Self Learning AI

2. Background

Cybersecurity is the most required field of division in this modern era, which solely runs on the digital media for data transfer and usage. In this day and age, thieves may not require your wallet to get your identification, but may readily steal it from the systems you are trusting. This unfolds the term “cyber-crime”. Traditional security measures face challenges due to evolving hacking techniques, including the use of script bots and AIs (Pearce et al., 2023), rendering manual penetration avoidance and security measures less effective.

Notably, the advent of AI has introduced new complexities, such as spear-phishing using GPT (Khan et al., 2021), overpowering some network-based intrusion detection systems (Gala et al., 2023). Researchers like Joshi et al. (2022), explored automatic penetration methods like SQL Injections, influencing our project's scope. Gallus et al. (2023) investigated how Generative Neural Networks, akin to DALL-E and ChatGPT, can serve as web application penetration testing tools, leveraging Large Language Models (LLM) and Generative Neural Networks.

Research by Li and Oprea (2016), on breach detection using security logs in an enterprise, inspired our AI-IDS to detect breaches through security logs. Memos and Psannis (2020) highlighted the use of AI in honeypots for botnet detection, a crucial component in countering Distributed Denial of Service (DDoS) attacks. Prasad et al. (2023), emphasized the role of GPT models in the cybersecurity domain, guiding our project's direction.