It copies alignment, not a label saying “good”
The important words are “to my own.” The ability uses the source’s alignment; it is not a hard-coded instruction that every button under a Good Cop display makes its target good.
Good Cop’s normal category is Innocent and normal alignment is good. The paired Bad Cop is normally an evil Underling that pretends to be Good Cop. The dated disguise mechanics explain why displayed function and actual alignment need separate treatment.
Resolve the source before choosing the target
Imagine two cards show Good Cop and you want to change a suspected target’s alignment. If you select the real good source, the conversion points toward good. If you use the evil counterpart, the direction is reversed. A plausible target does not make an unresolved source safe.
Look for a clue distinguishing the two actual identities or alignments before treating the ability as a rescue. Read the current case for any other conversion affecting the source itself; the normal reference label is not an immutable state.
Update alignment without inventing a new identity
The dated mechanics separate an alignment change from category and exact role identity. Record the target’s new alignment and the time of the change, keeping its existing identity question separate.
An older announcement about that target may describe the earlier state. Do not assume every previously displayed clue refreshes just because you used the ability. Seek information whose trigger actually checks the updated state.
The ordinary role descriptions say Good Cop and Bad Cop are paired. That is the supported baseline here, not a promise to resolve every special case-generation or chaos-ticket interaction.