What can Large Language Models do in Cyber Attacks on IoT device : Ethic hacking with ChatGPT towards Smart Cars ; Vad stora språkmodeller kan göra i cyberattacker på IoT-enheter : Etisk hackning med ChatGPT mot smarta bilar

Since the emergence of ChatGPT, large language models (LLM) have significantly outperformed traditional natural language processing models in complex language task processing and general task capabilities, bringing opportunities and challenges, such as in network attacks. This paper explores the application of LLMs in cybersecurity, with a special focus on IoT devices and smart cars. It takes an in-depth look at the capabilities of prominent LLMs, ChatGPT in the field of ethical hacking and its overall impact on cybersecurity. The study involved conducting a total of 8 tests, including 5 direct attack tasks, 2 system analysis tasks and reconnaissance tasks, on devices such as Ismartgate Pro and AutoPi to evaluate how helpful LLMs are in various hacking scenarios. The research results show that, with appropriate guidance, ChatGPT can significantly assist in all tasks including 5 attack tasks, including teaching basic knowledge, collecting information, and even formulating attack strategies and tool explanations, and was achieved in 3 attack tasks The script code for a successful attack. The study shows that LLMs have the potential to directly participate in attacks if improperly directed, giving them the potential to simplify the execution of cyberattacks and lower barriers for attackers with limited programming skills, further worsening the security of IoT devices. situation. This highlights the double-edged nature of this type of technology in terms of cybersecurity. ; Sedan uppkomsten av ChatGPT har stora språkmodeller (LLM) avsevärt överträffat traditionella bearbetningsmodeller för naturligt språk i komplex språkuppgiftsbearbetning och allmänna uppgiftskapaciteter, vilket medför möjligheter och utmaningar, till exempel i nätverksattacker. Denna artikel utforskar tillämpningen av LLM inom cybersäkerhet, med särskilt fokus på IoT-enheter och smarta bilar. Den tar en djupgående titt på kapaciteten hos framstående LLM:er, ChatGPT inom området etisk hacking och dess övergripande inverkan på ...