The rise of Generative Artificial Intelligence (GenAI) poses a fundamental challenge to education: how can we confidently assess whether students have truly learned when GenAI moves toward the ability of producing perfect outputs? This talk delves into the potential pitfalls of "deepfake learning" in the GenAI era, where traditional assessments struggle to distinguish between the learner’s use of GenAI and genuine learning. It then explores a shift in paradigms, proposing alternative assessment practices that prioritize the underlying cognitive processes of learning over mere output. From performance-based tasks and continuous feedback to collaborative assessments and metacognitive awareness, the talk showcases innovative methods for ensuring meaningful learning in a world with GenAI. Importantly, it also reframes GenAI as a potential tool for enhanced learning, offering personalized feedback, adaptive content, and engaging learning environments. Ultimately, the talk advocates for a balanced approach that leverages GenAI's strengths while employing robust assessment strategies that promote critical thinking and authentic learning. This discussion invites educators, technologists, and policymakers to collaborate in navigating the changing landscape of learning in the GenAI age. Be warned, there may be more questions presented in this talk than answers.