A customer complaint about contaminated food at Aldi escalated into a brand reputation issue when the retailer's response included visible AI prompt instructions in the reply itself. Jo Reeves discovered multiple hairs in a chicken caesar dip and contacted Aldi; the company's response, whilst offering refund guidance, exposed its underlying AI generation process by including the instruction "Make it short and concise but do not overdo it"—language clearly intended for a language model rather than a customer. Aldi subsequently acknowledged the response "fell well below our normal standards" and contacted Reeves directly to apologise, but the damage to perceived authenticity had already occurred. Reeves's frustration centred not on the product failure itself, which she acknowledged as human error, but on what she interpreted as inauthentic corporate communication masked by automation.
The incident exposes a critical tension in contemporary CX operations: the gap between deploying AI for efficiency and maintaining the brand voice customers expect. When support teams implement generative AI without adequate guardrails—particularly in complaint handling—they risk exposing the mechanical nature of their responses, which paradoxically undermines trust more than a delayed human reply would. This raises an uncomfortable question for teams already embedded in AI-assisted workflows: at what point does visible automation become a liability rather than an asset? Aldi's failure wasn't the use of AI itself, but the failure to sanitise outputs before delivery, suggesting that organisations rushing to scale AI-driven support without proper quality gates are trading short-term efficiency for long-term brand credibility.
For CX leaders, the lesson is unambiguous. Implementing AI in complaint resolution requires not just technical integration but rigorous output validation, particularly in high-stakes scenarios involving product safety or customer dissatisfaction. The reputational cost of appearing to automate empathy—or worse, to automate dismissal—outweighs the labour savings. Teams should audit their current AI implementations to identify whether prompt instructions, model artifacts, or tonally inconsistent language are leaking into customer-facing communications. Aldi's recovery through direct human contact suggests that transparency about AI use, paired with genuine resolution, may ultimately prove more defensible than attempting to hide automation altogether.
Woman shocked by Aldi's AI response to complaint BBC
Woman shocked by Aldi's AI response to complaint BBC