How could such systems prevent purposeful declining/refusal of correct answers, followed (within the "chat") by demand for "corrections" in misleading ways, and

How could such systems prevent purposeful declining/refusal of correct answers, followed (within the "chat") by demand for "corrections" in misleading ways, and stopping the "chat" only when the answer is obviously wrong? Couldn't instructions in one such solution, meant to DoS (not in volume, but in malicious purposely constructed "conversations") the competitor, lead to an overall degradation of quality in all, eventually?