Semek: Local AI hub for labelling/filtering digital content (itdaily.com)

Aaltra and Professor Ruben Verborgh of UGent have introduced Semek, a prototype AI device designed to help users control the digital content that enters their homes and workplaces. By processing AI requests locally, Semek empowers individuals to customize their content preferences, providing an alternative to traditional, often opaque filtering methods. As concerns rise over misinformation, fake imagery, and phishing attempts, Semek addresses these issues by allowing users to decide which content is relevant and safe, thereby reclaiming autonomy over their digital environment. The innovation behind Semek lies in its local processing capabilities, utilizing a Neural Processing Unit (NPU) that enables AI models to operate directly on the device. This means that personal data remains entirely within the user's control, enhancing privacy and reducing reliance on major technology firms. Users can configure the device to automatically label, blur, or remove unwanted content, ensuring a tailored experience without third-party interventions. Semek represents a significant shift towards user-centric AI systems, fostering greater security and independence while navigating the ever-increasing influx of digital information.