Data quality is deteriorating for some investment firms as artificial intelligence fuels a surge in vendors and changes how datasets are collected, processed and sold. A survey by consultancy Neudata found that one-third of investors who buy data said quality had worsened over the past couple of years. Hedge funds have long used alternative data—information outside traditional sources such as SEC filings and exchange trading data—to inform investment decisions. Daniel Entrup, cofounder of AggKnowledge, said providers are using AI either to generate or map data or to justify reducing or reallocating staff responsible for quality control. Buyers also say AI-generated products can be difficult to audit, creating compliance concerns around information availability and privacy, while the large language model (AI system trained to process and generate text) used to present the data may itself be unreliable. Hedge funds generally prefer to process raw data through their own systems, both because their internal methods can be part of their competitive advantage and because they trust AI more for workflow tasks than for investment signals. Entrup said basic data-cleaning problems have increased as vendors focus on new AI applications, while buyers have not seen a corresponding commercial benefit. A single hallucination—an AI-generated error presented as fact—can undermine an entire dataset and invalidate a signal.