ObjectiveTo review advances in liquid biopsy and artificial intelligence (AI) on early diagnosis of hepatocellular carcinoma (HCC). MethodsRecent studies on early diagnosis of HCC were summarized. ResultsTraditional biomarkers has limited sensitivity for early diagnosis of HCC, whereas combined biomarker-based strategies can improve diagnostic performance. Circulating tumor DNA, circulating tumor cells, exosomal RNA and protein biomarkers can compensate for conventional detect approaches. Multi-analyte panels show superior diagnostic performance compared than single detection agent. AI-based approaches can support the identification and selection of key molecular features, the integration of multi-source data, and risk stratification, thereby improving the detection of early-stage HCC. However, their clinical implementation remains limited by the lack of unified detection workflows, insufficient multicenter external validation, and inadequate model interpretability. ConclusionsEarly diagnosis of HCC is moving from single serum biomarkers toward to combined markers, and further progress to AI-assisted stratified screening. In future, multicenter prospective validation, build standardized workflows and biologically interpretable models can improve early diagnosis of HCC.