ODRL is becoming the standard way to express who may use what data, under which conditions, across European dataspaces and AI governance. The standard specifies how to write policies. It does not specify, mathematically, how a system should decide whether a given request satisfies one.

That omission has predictable consequences. Every implementation supplies its own interpretation, so the same policy can be permitted by one system and refused by another, and no tool can guarantee it agrees with any other. For a language whose entire purpose is expressing rules that hold across organisations, that is a serious gap.

This work provides a formally grounded evaluator and compares it against existing implementations. The comparison is where the value lies: it makes the divergences visible and specific, rather than leaving them as a suspicion that different tools probably disagree somewhere.