American students use AI agents to automate entire online courses

As AI agent models proliferate, some American students have moved beyond using chatbots to look up answers and are now delegating entire online courses to AI agents. The agents operate behind the scenes, completing course tasks while the student attends to other activities. The phenomenon exposes a significant vulnerability in online educational platforms regarding fraud detection compared to in-person classes. Most online education activities occur on screens, allowing AI agents to log into platforms like Canvas and Blackboard to take tests, watch course videos, write papers, and even participate in discussions on behalf of the student. AI companies including OpenAI (ChatGPT), Google (Gemini), and Grammarly have implemented virtually no measures to prevent these uses and have not blocked AI agents from learning platforms or responded to educators' requests that models identify themselves when accessing these systems. Perplexity, which develops AI-powered browsers, claimed that requiring agent identification would jeopardize student privacy and security—a justification educators have questioned. Universities are responding with varied policies. The University of Chicago incorporated a prohibition on phones, tablets, and computers in classrooms as part of its new AI strategy for first-year courses. Princeton University discontinued its century-old honor code that had allowed unsupervised testing following an AI fraud incident.