AIproposesCodeVerifies v1.0.0: Local Rule-Based Checker for Data Preservation in AI-Rewritten Text

Abstract

AIproposesCodeVerifies is a local tool that checks whether a text rewritten or translated by an AI assistant preserves the data the user has chosen to protect. Principle: the AI proposes, the code verifies, the person decides. The tool compares the original text with the AI's proposal and detects certain classes of changes, omissions and additions in dates, times, amounts, deadlines, legal references, URLs, Spanish national ID numbers (DNI/NIE), and user-defined protected, required or forbidden terms. It also applies a small set of optional style rules. Users define their checks through exportable and importable JSON profiles. Everything runs in the browser. There is no server, no network communication, and no machine-learning model: the checks are deterministic, rule-based code. The user copies text to and from the AI service manually; the tool does not connect to any AI service. Package contents (version 1.0.0): - index.html: self-contained web version. - extension/: Chrome/Edge extension (Manifest V3, side panel and context menu; permissions: sidePanel, storage, contextMenus; no host permissions). - engine/: verification engine (v12) with SHA-256 integrity check. - tests/ and scripts/: regression, profile and packaging tests. - README, LICENSE and CITATION.cff. Scope and limits: the tool detects certain classes of errors; it does not check meaning, factual truth, or whether a cited legal norm is correct, and it does not guarantee that an AI output is correct. The engine is in a validation phase and has been developed and tested with Spanish-language texts. No external user study has been conducted for this release. Privacy: texts checked by the user are not sent to any server. Profiles are stored locally in the browser; a temporary working draft is kept in session storage.