Abstract
This paper investigates the systemic opacity and semantic vagueness characterizing environmental claims in the non-financial disclosures of major European banking institutions. By cross-referencing the emerging European legal framework – in particular after the introduction of the new Directive 2024/825 – with empirical findings obtained through algorithmic analysis of consolidated sustainability reports, the paper exposes widespread use of vague, unverifiable, and prospective ESG statements that could qualify as misleading under EU law. Using a Python-based natural language processing tool, the study identifies recurrent linguistic patterns that diverge from the transparency, verifiability, and measurability standards now mandated. It argues for the adoption of algorithmic auditing both as a preventive compliance mechanism and as a regulatory enforcement tool, proposing a shift towards a new model of semantic supervision capable of aligning rhetorical ESG narratives with demonstrable corporate commitments.