Deep learning enables the discovery of antibiotic peptides through molecular de-extinction. Researchers used deep learning models to analyze the proteomes of extinct organisms, identifying 37,176 peptides with broad-spectrum antimicrobial activity, 11,035 of which were not found in extant organisms. They synthesized 69 peptides and confirmed their activity against bacterial pathogens. Most peptides killed bacteria by depolarizing their cytoplasmic membrane, unlike known antimicrobial peptides that target the outer membrane. Lead compounds, including mammuthusin-2 from the woolly mammoth, elephasin-2 from the straight-tusked elephant, hydrodamin-1 from the ancient sea cow, mylodonin-2 from the giant sloth, and megalocerin-1 from the extinct giant elk, showed anti-infective activity in mice with skin abscess or thigh infections. Molecular de-extinction aided by deep learning may accelerate the discovery of therapeutic molecules. Antimicrobial-resistant infections cause approximately 1.27 million deaths annually, with projections of 10 million by 2050. The World Health Organization estimates that 24 million individuals could face extreme poverty due to the high cost of treating these infections by 2030. Molecules serve as records of evolutionary history and may provide blueprints for therapeutic design. Molecular de-extinction refers to the resurrection of extinct molecules to tackle contemporary challenges such as antibiotic resistance. By uncovering a new sequence space of previously unexplored molecules, molecular de-extinction offers a promising approach to expand our vision of life's molecular diversity while helping unveil molecules that may play a role in host immunity throughout evolution. Molecular de-extinction has already yielded preclinical antibiotic candidates such as neanderthalin-1 (A0A343EQH4-LAM11). Recent computational and artificial intelligence approaches have been developed for antibiotic discovery. Machine-learning models have been used to generate antibiotics and predict antimicrobial activity, haemolysis, and antimicrobial resistance. Computational methods have been developed to discover new antibiotics through proteome mining. We previously mined the human proteome as a source of antibiotics and identified encrypted peptides (EPs), fragments within proteins that possess antimicrobial properties. We hypothesized that EPs exist not only in modern humans but also throughout evolution. Thus, through paleoproteome mining and ML, we identified similar molecules in ancient humans. These recent computational efforts have greatly accelerated our ability to discover new preclinical antibiotic candidates. In this article, we introduce antibiotic peptide de-extinction (APEX), a new multitask deep learning approach used to systematically mine all available proteomes of extinct organisms for antibiotic discovery. APEX was trained on peptide data from both our in-house dataset and the publicly available Database of Antimicrobial Activity and Structure of Peptides (DBAASP). APEX utilizes an encoder neural network, combining recurrent and attentionDeep learning enables the discovery of antibiotic peptides through molecular de-extinction. Researchers used deep learning models to analyze the proteomes of extinct organisms, identifying 37,176 peptides with broad-spectrum antimicrobial activity, 11,035 of which were not found in extant organisms. They synthesized 69 peptides and confirmed their activity against bacterial pathogens. Most peptides killed bacteria by depolarizing their cytoplasmic membrane, unlike known antimicrobial peptides that target the outer membrane. Lead compounds, including mammuthusin-2 from the woolly mammoth, elephasin-2 from the straight-tusked elephant, hydrodamin-1 from the ancient sea cow, mylodonin-2 from the giant sloth, and megalocerin-1 from the extinct giant elk, showed anti-infective activity in mice with skin abscess or thigh infections. Molecular de-extinction aided by deep learning may accelerate the discovery of therapeutic molecules. Antimicrobial-resistant infections cause approximately 1.27 million deaths annually, with projections of 10 million by 2050. The World Health Organization estimates that 24 million individuals could face extreme poverty due to the high cost of treating these infections by 2030. Molecules serve as records of evolutionary history and may provide blueprints for therapeutic design. Molecular de-extinction refers to the resurrection of extinct molecules to tackle contemporary challenges such as antibiotic resistance. By uncovering a new sequence space of previously unexplored molecules, molecular de-extinction offers a promising approach to expand our vision of life's molecular diversity while helping unveil molecules that may play a role in host immunity throughout evolution. Molecular de-extinction has already yielded preclinical antibiotic candidates such as neanderthalin-1 (A0A343EQH4-LAM11). Recent computational and artificial intelligence approaches have been developed for antibiotic discovery. Machine-learning models have been used to generate antibiotics and predict antimicrobial activity, haemolysis, and antimicrobial resistance. Computational methods have been developed to discover new antibiotics through proteome mining. We previously mined the human proteome as a source of antibiotics and identified encrypted peptides (EPs), fragments within proteins that possess antimicrobial properties. We hypothesized that EPs exist not only in modern humans but also throughout evolution. Thus, through paleoproteome mining and ML, we identified similar molecules in ancient humans. These recent computational efforts have greatly accelerated our ability to discover new preclinical antibiotic candidates. In this article, we introduce antibiotic peptide de-extinction (APEX), a new multitask deep learning approach used to systematically mine all available proteomes of extinct organisms for antibiotic discovery. APEX was trained on peptide data from both our in-house dataset and the publicly available Database of Antimicrobial Activity and Structure of Peptides (DBAASP). APEX utilizes an encoder neural network, combining recurrent and attention