PH student participates in building AI that uses human genome to generate and protect digital credentials

As digital systems continue to handle more sensitive information, the demand for stronger and more reliable security mechanisms is increasing. Researchers are now exploring alternatives to traditional cryptographic approaches to address limitations in entropy, reuse, and long-term resilience. One emerging direction combines artificial intelligence with biological data such as the human genome, leading to new methods for generating and protecting digital credentials.

DNAcrypt-AI explores the use of the human genome as a source for generating high-entropy digital credentials such as passwords and cryptographic keys [1]. Unlike conventional systems that rely on random number generators or reusable secrets, genomic data provides a highly complex and diverse information space. When combined with artificial intelligence, this approach allows the generation, encryption, and decryption of security keys with significantly higher entropy, making them more resistant to brute-force attacks and credential reuse.

According to project leader Marvin De los Santos of ChordexBio, DNAcrypt-AI was developed to explore how biological data and AI can be responsibly applied to modern security problems. “DNAcrypt-AI is an experimental AI system that studies how genomic patterns can be encoded and transformed into cryptographic materials,” he explained. “What excites us most is not just the technical side, but the possibility of creating security systems that are both scientifically grounded and adaptable to future digital demands.”

Building the system came with several challenges, particularly in translating complex genomic representations into formats suitable for cryptographic modeling. Through the assemBold program, Ms. Lynn, a participating student, worked closely with the research team to address these challenges. Structured mentorship, guided experimentation, and access to real research workflows allowed her to contribute meaningfully despite the technical difficulty of the project. The program emphasized learning by building, rather than following pre-made pipelines.

Looking ahead, the DNAcrypt-AI team sees genome-powered cryptography as a potential response to the increasing need for secure digital identities, credentials, and authentication systems. As societies become more dependent on digital platforms for healthcare, finance, and communication, traditional security models may no longer be sufficient. AI-driven cryptography using complex biological data offers one possible direction for addressing these emerging risks.

The assemBold program continues to support similar projects where students and early-career researchers can participate in real scientific development. Beyond building future technologies, the program focuses on creating solutions to present-day problems using published-first, scientifically rigorous approaches. Through assemBold, participants are encouraged not only to learn advanced skills, but also to contribute to research that has clear relevance to society.

Users can test and validate DNAcrypt-AI. Protocols are already available for generating and encrypting passwords [2] and secret keys [3], including decrypting credetials [4].

References:

[1] DNAcrypt-AI (GitHub). https://github.com/mahvin92/DNAcrypt-AI

[2] DNAcrypt-AI protocol for generating and encrypting a password. DOI: 10.17504/protocols.io.bp2l6e951gqe/v1

[3] DNAcrypt-AI protocol for generating and encrypting a secret key. DOI: 10.17504/protocols.io.14egn12k6v5d/v1

[4] Decryption of genome-encoded cryptographic keys using DNAcrypt-AI. DOI: 10.17504/protocols.io.5qpvo1387g4o/v1

Ways to join assemBold:

Have a specific topic →

Join partner programs →

Participate in an open research →