AI-designed phages illustrate the AIxBio pipeline’s potential and risks, underscoring the need for careful policy development and responsible deployment as the field matures.

A new AI BioDesign accelerator unites the Allen Institute, University of Washington, and Fred Hutch Cancer Center to study how nature designs biological systems using artificial intelligence, large-scale experiments, and open science.

Funding comes from FFST, totaling $94.6 million over five years, with allocations of roughly $46.1 million to the Allen Institute, $43.8 million to UW, and $4.7 million to Fred Hutch.

The initiative blends open-science platforms, genome science, translational medicine, and cellular systems to create a continuous design-build-measure-learn loop across the collaboration.

Public statements emphasize accelerating AI-driven experimentation to inform biological design and move toward more predictable engineering of living systems.

Safeguards include excluding eukaryotic viruses from training data, responsible biodesign principles, and ongoing discussions about DNA synthesis screening and robust, structure- and function-based screening beyond sequence similarity.

Initial targets include custom disease-binding proteins, genetic switches to regulate gene expression, and tools to destroy or stabilize specific proteins to extend capabilities beyond natural evolution.

As cloud labs become more accessible, governance must address the balance between rapid innovation and safety with policies and oversight.

Goals encompass advancing therapies, environmental applications like plastic-dissolving proteins, and programmable cells to shorten the cycle from concept to testing.

Team aims to show real progress in 18–24 months and to broaden global access to tools and results within five years.

Stanford-Arc’s Evo 2 model designed phages; 285 designs were synthesized and 16-phage cocktails restored activity against resistant E. coli, illustrating AI-designed phage viability as a proof of concept.

David Baker leads AI BioDesign as scientific director, signaling a shift from asking what nature has already made to exploring what AI-designed biology can test and achieve.