<h1>AI customer-contact tool claims 90%+ accuracy and 88% resolution - liability, third-party LLM reliance, data privacy risks</h1> A vendor announced that its customer-contact AI achieved reported response accuracy above 90% and an independent resolution rate of 88% following user testing, attributing results to combining generative models, retrieval-augmented generation, and smaller task-specific models, while relying on multiple third-party LLM providers. Legally relevant issues include potential liability for erroneous or hallucinated responses despite claimed mitigations, third-party provider dependence and vendor representations that could trigger warranty or advertising scrutiny, and data-processing, privacy and transparency obligations (including model provenance, training data use and cross-border transfers) under applicable consumer protection and data-protection laws.