Chainalysis said machine learning should not be treated as definitive evidence in blockchain investigations, arguing that predictive models can help spot patterns in large volumes of on-chain data but are not suitable for foundational claims about wallet control. In a report published August 14, the blockchain analytics firm said it does not use machine learning to identify wallet segments or determine whether multiple blockchain addresses are controlled by the same entity. It classifies wallet clustering as Tier 1 structural intelligence, meaning the methodology must be deterministic, reproducible and auditable. The firm said AI remains useful for lead generation, anomaly detection, pattern recognition and evidence-based category assessments, provided those outputs are treated as probabilistic signals and undergo additional validation. Chainalysis also linked the issue to legal scrutiny under the Daubert standard (U.S. test for expert evidence), citing the 2024 United States v. Sterlingov case, where it said a court found its clustering methodology sufficiently sound after reviewing how clusters were built and whether the reasoning could be independently verified.