Adaptive feedback control of neuronal activity is a powerful technique for both revealing and repairing neuronal function, but our knowledge on its effective implementation is still very limited. Presently, electrical brain stimulation, a well-stablished therapeutic to address neurological disorders such as Parkinson’s disease and epilepsy, typically relies on open-loop protocols. However, electrical stimulation should adapt to the neuronal activity and control the pathological dynamics with minimal perturbations. Here we tested for the first time in neuronal populations (in vitro), a popular control technique used in multiple computational studies to disrupt pathological neuronal oscillations - Delayed Feedback Control (DFC). We show that DFC may worsen the oscillatory behaviour, promoting faster bursting. Alternatively, we developed and validated an alternative method, adaptive DFC (aDFC), which monitors the ongoing periodicity and self-tunes accordingly. aDFC disrupts the collective neuronal oscillation and leads to a decrease in network synchrony, making it a better candidate for therapeutic neurostimulation.