Firmware attacks pose significant security risks to IoT devices due to often insufficient security protection. Such attacks can result in data breaches, privacy violations, and operational disruptions. For example, attackers can alter the structure of packets (e.g., header, payload) to create malformed packets and send them to nodes in a victim network. Those packets then remotely trigger vulnerabilities (e.g., buffer overflows) present in the firmware, potentially resulting in crashes or undefined behavior. In this paper, we first investigate how firmware vulnerabilities (e.g., out-of-bounds errors) can be remotely exploited in 6LoWPAN-based IoT networks by analyzing bug reports from IoT projects. Then, we introduce a protocol fuzzing framework that leverages the GPT-4 model to guide the creation of test cases as malformed packets. Later, these packets are sent to nodes to detect potential vulnerabilities. For evaluation, we build a testing network composed of real IoT devices to execute the test cases. As a result, we discovered four new vulnerabilities, including three cases of uncontrolled resource consumption and one issue of improper input validation, which can be remotely exploited in the Contiki-NG operating system.