Devices are frequently targeted by attacks, posing considerable risks to smart homes. The lack of standardisation in security updates makes it difficult to protect smart homes, making them potentially vulnerable. Additionally, updates are usually entirely dependent on actions performed statically by users. Finally, IoT manufacturers commonly abandon their products after launch and no longer support security updates. In this context, this paper presents a solution that supports the dynamic management of smart home security parameters with the aid of humans, i.e., the so-called Human-In-the-Loop (HIL). The solution uses autonomous computing techniques to guarantee the self-protection of home devices with the possibility of human assistance in the most critical decisions. Combining the human-in-the-loop approach and autonomous computing strengthens smart home security, offering an adaptive and effective defence against constantly evolving vulnerabilities. The results of experiments show the solution’s effectiveness in mitigating vulnerabilities at different levels of criticality.