AI-driven IoT ecosystems are pivotal to technological advancements, offering enhanced efficiency and functionality across various sectors. However, integrating AI into IoT devices presents energy efficiency, security, privacy, and sustainability challenges. Addressing these issues is crucial to align IoT systems with global sustainability goals. This study investigates and optimizes AI algorithms in IoT environments to foster secure, energy-efficient, and sustainable ecosystems. The research includes a systematic review of AI algorithms suitable for IoT applications and the development of performance metrics evaluating efficiency, scalability, and sustainability. Adaptive, context-aware AI techniques were explored to optimize energy use, while security and privacy challenges were analyzed with mitigation strategies. An environmental impact framework assessed practices for minimizing electronic waste and energy consumption. Findings reveal that traditional supervised and unsupervised learning algorithms exhibit robust efficiency but struggle with scalability and real-time adaptability. Advanced techniques like reinforcement learning and deep learning show superior performance but demand significant computational resources. Adaptive AI techniques effectively optimize energy consumption dynamically, enhancing efficiency. Security and privacy analyses uncovered vulnerabilities, with strategies proposed to strengthen trust. Environmental assessments highlighted the need for durable IoT designs, eco-friendly materials, and energy optimization. The study underscores the importance of lightweight, scalable AI algorithms tailored for IoT to balance performance and energy efficiency. Enhancing security frameworks and incorporating sustainability practices in IoT design and manufacturing are essential. Future research should focus on real-world validation, refined algorithms, and advancing sustainability practices to ensure responsible digital transformation.