A recent study conducted by researchers at the University of Oslo reveals how artificial intelligence is being integrated into lower secondary education across Norway. Led by Ph.D. candidate Annie Karoline Elgen alongside colleagues Katherina Dodou and project leader Professor Lisbeth M. Brevik, the EDUCATE research initiative analyzed video recordings from 75 lessons across 20 classrooms. The findings indicate that AI was actively utilized in 18 percent of observed classes, with educators serving as the primary drivers of integration. The peer-reviewed results were published in Teaching and Teacher Education. Rather than deploying AI as a standalone tool, teachers strategically combined subject instruction with generative AI literacy. In English and literature classes, educators assigned tasks such as feeding student-written scripts into AI models for feedback, simplifying complex historical texts for accessibility, and prompting chatbots to simulate literary characters for deeper thematic analysis. A recurring pedagogical focus was critical evaluation. Instructors explicitly cautioned students that AI lacks critical reasoning capabilities and can generate both accurate and misleading information. Teachers emphasized that students must possess sufficient foundational knowledge to independently verify AI outputs, ensuring that digital tools complement rather than replace academic rigor. Despite structured guidance, the study uncovered significant challenges regarding student comprehension and usage boundaries. Researchers observed instances where learners misinterpreted instructions to use AI merely for inspiration, instead submitting machine-generated responses as their own work. This pattern highlights a broader educational gap: students often struggle to distinguish between appropriate AI-assisted learning and academic dependency. The researchers note that while digital tools offer substantial pedagogical advantages, unguided reliance on language models risks undermining the development of essential reading, writing, and analytical skills. To address these inconsistencies, the research team advocates for standardized, locally defined policies governing AI use in educational settings. Current guidelines vary widely across Norwegian institutions, leaving both educators and students navigating uncertain expectations. The researchers align with a recent UNICEF report noting that students themselves find existing regulations ambiguous and request transparent, school-specific frameworks. Successful implementation, they argue, requires teachers to possess both deep subject expertise and robust digital competency, enabling them to navigate emerging technologies while safeguarding academic integrity. As artificial intelligence continues to reshape classroom dynamics, the study underscores the necessity of deliberate, critically informed integration. Educators who successfully merge AI instruction with traditional curricula demonstrate that the technology value lies not in automation, but in fostering analytical competence and responsible digital citizenship.