The evolutionary birth of proteins is essential for the diversity of life, but typically research focuses on de novo transcript evolution with little attention to the biochemical steps that enable completely new proteins to evolve. Sunflower PawS1 is an unusual albumin precursor because in addition to albumin it yields a potent, bicyclic protease-inhibitor. We have traced the evolutionary origin of this inhibitor peptide and showed it evolved naturally in steps starting 45 million years ago. We also show using synthetic constructs encoding protein chimeras that the inhibitor peptide is capable of emerging from alternative locations within its native host as well as from unrelated seed proteins when they host it. This dominance over host proteins demonstrate the inhibitor peptide is a hijacker rather than a passive passenger to albumin within PawS1. This de novo protein evolution inside a benign region of a pre-existing protein bypasses many impediments de novo evolved genes face before encoding a stable, new protein. Our current efforts are to characterise a separate but potentially similar and ancient sequence hijacking and to determine if hijacking is a widespread protein evolutionary phenomenon.