Scientists Develop Molecular Watermark to Identify AI-Designed Proteins
Boston: Scientists have developed a new approach that could help researchers identify proteins designed with artificial intelligence by introducing a molecular-level signature that can be detected through laboratory analysis.

The research addresses an emerging challenge in biotechnology: as AI systems become capable of designing new proteins, researchers need reliable ways to determine whether a biological molecule was created using computational design.
AI is changing protein research
Artificial intelligence has rapidly expanded its role in structural biology and protein engineering.
Modern AI systems can analyse enormous biological datasets and generate designs for proteins with specific characteristics. Researchers are exploring these capabilities for applications including medicine, industrial biotechnology and materials science.
However, the growing use of AI also creates questions about traceability.
A protein created through computational design may be difficult to distinguish from one developed through conventional laboratory methods once it exists as a physical molecule.
A molecular signature
Researchers have proposed a technique that effectively creates a molecular watermark.
The concept involves incorporating a detectable molecular feature into an AI-designed protein so that scientists can later determine whether the protein originated from a particular design process.
The watermark is intended to remain identifiable through laboratory testing without fundamentally changing the purpose of the protein.
Why traceability matters
Being able to determine the origin of biological designs could become increasingly important as AI-generated proteins enter research laboratories and potentially commercial applications.
Traceability could help researchers document the development history of engineered proteins and distinguish between naturally occurring biological molecules and computationally designed ones.
It could also support laboratory safety and responsible biotechnology practices.
Potential role in biosecurity
AI-assisted protein design has attracted attention because the same technology can potentially be used for beneficial or harmful purposes.
Researchers are therefore examining ways to improve oversight as computational biology becomes more powerful.
A molecular watermark could provide one possible mechanism for tracking the origin of engineered biological materials.
However, such technologies are only one part of a broader safety framework.
Challenges remain
Molecular tagging must be reliable enough to survive protein production and laboratory processing.
Scientists also need to determine whether the watermark can be removed, altered or unintentionally lost during subsequent experiments.
Another challenge is ensuring that the tagging method does not interfere with the biological activity or structural properties of the protein being studied.
AI and biotechnology converge
The research illustrates how artificial intelligence is moving beyond software and into the physical sciences.
AI systems can now assist scientists with the design of molecules and proteins that may eventually be produced and tested in laboratories.
This creates new opportunities for drug discovery and biotechnology while also increasing the need for methods that establish provenance and maintain scientific traceability.
Future applications
If molecular watermarking can be developed into a reliable and broadly applicable system, it could potentially be used alongside other laboratory authentication and tracking methods.
Researchers may also explore whether similar approaches could be applied to different types of engineered biological molecules.
The development reflects a growing effort to ensure that advances in AI-driven biotechnology are accompanied by tools that support transparency, verification and responsible research.
As AI-generated biological designs become more common, identifying where a molecule came from could become nearly as important as understanding what the molecule does.