Objectives:
The article asks whether Polish and European Union evidence law can secure the reliability of evidence that may have been fabricated with generative artificial intelligence (GenAI)—from deepfakes to synthetic documents and electronic signatures.

Material and methods:
It applies the legal-dogmatic method to the Polish procedural codes and EU law, using the United States as a heuristic reference point rather than a full comparator and analysing the European case law binding on Poland.

Results:
Polish law has no dedicated regime for electronic or AI-generated evidence: such material is admitted and assessed under the free-evaluation principle, with authentication delegated to a court-appointed expert (biegły). EU law adds two functional layers—the eIDAS trust-service presumptions and equivalence rules (confined to the cryptographic container, not the content) and the AI Act’s marking duties (which fabricators will not observe).

Conclusions:
The continental model relocates rather than removes the gatekeeping problem; the fair-trial right to challenge evidence (Article 6 ECHR) is the decisive safeguard. De lege ferenda proposals for Poland and the EU are advanced.