Criminals are exploiting AI-powered search tools and chatbots to inject fraudulent customer service phone numbers into search results, directing consumers away from legitimate support channels and towards scam operations. Virgin Media O2's research has documented actual fraud cases where customers, believing they were contacting official support lines, instead reached criminals who harvested personal data and financial information. The attack vector is particularly effective because it exploits the moment of highest customer vulnerability—when someone is actively seeking help and trusting the first result returned by an AI tool or search engine.
For CX teams, this represents a fundamental threat to the trust relationship between brands and customers. When your carefully designed support infrastructure—whether built on Zendesk, Freshdesk, or Salesforce Service Cloud—is undermined by fraudsters poisoning the discovery layer, customers lose confidence in the entire support ecosystem. The question becomes acute for organisations deploying AI-powered chatbots and voice agents: how do you ensure your own AI implementations aren't inadvertently legitimising the appearance of fake numbers through their responses? Teams must now treat phone number verification as a critical security control, not merely a routing detail, and audit whether their knowledge bases and AI training data could be weaponised to direct customers elsewhere.
The broader implication is that CX professionals can no longer treat customer authentication and channel verification as separate from their support strategy. Support team leads need to implement verification protocols that assume customers may have been directed through compromised channels, whilst administrators should audit their systems for how phone numbers and contact information are surfaced—both internally and through any public-facing AI tools. This is no longer a fraud prevention problem delegated to security teams; it's a customer experience design problem that sits squarely in the CX function's remit.
New fraud warning after borough residenst served fake customer service phone numbers by AI Gedling Eye