This article introduces a deep learning-based approach called Antibiotic Peptide De-extinction (APEX) to discover new antibiotics from the proteomes of extinct organisms. APEX uses a multitask deep learning model to predict the antimicrobial activity of peptides, identifying 37,176 sequences with broad-spectrum antimicrobial activity, 11,035 of which are not found in extant organisms. The study synthesized and experimentally validated 69 peptides, finding that most killed bacteria by depolarizing their cytoplasmic membrane. Notably, several lead compounds, including mammothusin-2 from the woolly mammoth and elephasin-2 from the straight-tusked elephant, showed anti-infective activity in mice with skin abscess or thigh infections. The results demonstrate that molecular de-extinction, aided by deep learning, can accelerate the discovery of therapeutic molecules to combat antibiotic resistance.This article introduces a deep learning-based approach called Antibiotic Peptide De-extinction (APEX) to discover new antibiotics from the proteomes of extinct organisms. APEX uses a multitask deep learning model to predict the antimicrobial activity of peptides, identifying 37,176 sequences with broad-spectrum antimicrobial activity, 11,035 of which are not found in extant organisms. The study synthesized and experimentally validated 69 peptides, finding that most killed bacteria by depolarizing their cytoplasmic membrane. Notably, several lead compounds, including mammothusin-2 from the woolly mammoth and elephasin-2 from the straight-tusked elephant, showed anti-infective activity in mice with skin abscess or thigh infections. The results demonstrate that molecular de-extinction, aided by deep learning, can accelerate the discovery of therapeutic molecules to combat antibiotic resistance.