Universities are confronting two simultaneous AI-era shifts: faculty are working to integrate large language models into teaching while others scramble for reliable ways to detect unauthorized AI use in student work. Some instructors are redesigning assessments and pedagogies to incorporate AI as a tool; others are pushing for technology and frameworks to identify AI-assisted submissions. Reporting and opinion pieces show instructors replacing traditional analytical papers with in-class exams, viva voce presentations and project formats where AI use is observable or explicitly permitted. Advocates for detection tools argue that reliable provenance methods would let faculty set clear policies: forbid AI where it undermines learning and permit it where it augments it. The debate engages campus IT, academic affairs, and legal teams. Clarification: “LLM” refers to large language models—AI systems trained on massive text corpora that can generate essays and answers. Administrators should prioritize assessment design, detection tools, and faculty training as complementary responses rather than relying on any single tactic.