The attribution crisis in LLM search results: Estimating ecosystem exploitation

Today's article comes from the Cambridge Data & Policy journal. The authors are Strauss et al., from University College London. In this paper, they're arguing that Large Language Models (LLMs) are perpetrating ecosystem-exploitation on a massive scale. That they're extracting value from external knowledge sources, synthesizing those inputs into outputs, profiting from those outputs, then failing to return visibility, traffic, or credit to the sources they used.

DOI: 10.1017/dap.2026.10064