A researcher is applying computational cognitive neuroscience to predict the failure modes of the first transformative AI. By modeling brainlike AGI, the author aims to identify efficient interventions for rushed development timelines. This mechanistic approach seeks to preempt alignment collapses. It offers a niche, academic perspective on safety rather than a scalable technical solution.