Abstract <p><i>Phaeocystis globosa</i> (<i>P. globosa</i>), a main culprit of harmful algal blooms (HABs), is highly prone to blocking the inlet filter screens of nuclear power cold sources, thus posing a significant threat to nuclear power safety. However, existing methods for <i>P. globosa</i> detection fail to achieve rapid and on-site monitoring of single-cell densities prior to bloom outbreaks, limiting timely defensive measures. In this study, we developed a novel biosensor platform for efficient <i>P. globosa</i> detection, leveraging an Exo III-assisted signal amplification strategy to significantly enhance sensitivity and selectivity. The biosensor achieved an ultra-low limit of detection (LOD) of 268.91 fg µL⁻¹ (3119 cells L⁻¹), far below the benchmark concentration for <i>P. globosa</i> blooms (10⁷ cells L⁻¹), and demonstrated a wide linear detection range from 500 fg µL⁻¹ to 10 ng µL⁻¹. Furthermore, the biosensor’s accuracy and reliability were validated through comparative analysis with droplet digital PCR (ddPCR) using actual samples from the Beibu Gulf of China, revealing a low risk of <i>P. globosa</i> blooms in the region at the sampling time. This study represents a significant advancement in HAB monitoring by providing a highly sensitive, rapid, and field-deployable tool for early warning of <i>P. globosa</i> blooms. The biosensor’s innovative design and performance address critical gaps in current detection methods, offering practical implications for safeguarding coastal nuclear power facilities and protecting marine ecosystems.</p> Graphical Abstract <p></p>