Generation of Programmatic Rules for Document Forgery Detection Using Large Language Models

Document forgery poses a growing threat to legal, economic, and governmental processes, requiring increasingly sophisticated verification mechanisms. Recent advances in code generation with large language models (LLMs) offer new potential for automating and scaling the generation of rule-based plausibility checks for forgery detection. This work investigates the extent to which LLMs, adapted on domain-specific code and data through different fine-tuning strategies, can generate executable and effective verification procedures for previously unseen forgery patterns, highlighting their potential as scalable tools to support human decision-making in security-sensitive contexts where comprehensibility is required.