August 6, 2026

expert reaction to generative design of bacteriophages with genome language models

A study published in Science looks at generative design of bacteriophages using genome language models. 

Prof Patrick Cai, Chair of Synthetic Genomics, Manchester Institute of Biotechnology, University of Manchester, said:

“This is an important milestone for synthetic genomics. For the first time, we are seeing AI move beyond predicting biological sequences to generating entire functional genomes that work in the laboratory. While these are relatively small bacteriophage genomes, the significance extends far beyond phages. It suggests that genome language models are beginning to learn the design principles encoded by evolution, opening the door to AI-assisted genome writing. The next challenge is to scale these approaches to much larger and more complex genomes while ensuring that every computational prediction is matched by rigorous experimental validation.”

Comments from our friends at SMC Spain:

Jordi García Ojalvo, Professor of Systems Biology, Pompeu Fabra University of Barcelona, said:

“The press release accurately reflects the content of the article, although it focuses heavily (from the very first sentence) on microbial resistance – which is only part of the findings – and on biosafety issues, which, again, are only part of the discussion. The breakthrough achieved is significant, as it demonstrates that a complete functional genome can be designed from scratch using generative AI, but there are a few points to consider.

“Firstly, the efficiency of the process is low: out of thousands of genomes generated, only 16 viable phages are obtained (it could be argued that the rest are ‘hallucinations’ of the model – at least the remaining 300 genomes tested in the laboratory, out of the thousands designed). It is to be hoped that future versions of this procedure will improve efficiency, but it is unclear whether a qualitative leap will be achieved, as the data available to these genomic language models is (and will remain) limited. This low efficiency is reminiscent of Yamanaka’s stem cell reprogramming, for which he was awarded the Nobel Prize in 2012; whilst efficiency has been gradually improving, 14 years on it remains low (in this case, with generative AI, it is more difficult to implement).

“Secondly, from a biosafety perspective, in my opinion the risk here is lower than with traditional LLMs, as the designed genomes must be tested in the laboratory one by one, as was done in the article, and, as I have said, the efficiency is low. It is difficult to imagine these models automatically generating viable genomes ‘out-of-the-box’.

“Finally, I would also like to emphasise that these advances do not help us understand why some genomes are viable and others are not. Understanding is a human capacity, and in this respect, generative AI cannot replace us”.

Marc Güell, coordinator of the Translational Synthetic Biology research group and ICREA research professor,  Pompeu Fabra University (UPF), said:

“The study is of a very high standard. The laboratory leading it is at the forefront of generative AI applied to biology. In my laboratory, Synbio Lab at the UPF in Barcelona, we are regular users of the generative AI tools developed by Hie and colleagues. Their DNA LLMs are the most advanced in the world.

“The experiments designed to demonstrate the results experimentally appear to be very well conducted.

“This is another step towards the bio-design capabilities of generative AI. We are witnessing a very significant turning point. For the first time in history, we are beginning to design biology on a computer. Until now, we could read and rewrite, but we could not generate new synthetic content. Thanks to AI, we are beginning to move away from relying exclusively on bioprospecting. It has been demonstrated that we can already design relatively small synthetic biological systems such as proteins (synthetic binders, synthetic Cas9, synthetic transposases or synthetic enzymes); here, they go a step further by designing an entire synthetic phage comprising more than ten proteins and a genome of thousands of bases. It is an exciting time. The study appears to be very robust. Being able to design biology on a computer allows us to dream of exciting possibilities for tackling humanity’s greatest challenges. Synthetic binders have been created that mimic the antibodies used in immunotherapy; for example, synthetic Cas9 enzymes are being applied to treat genetic disorders, whilst synthetic phages could lead to potential antibiotic therapies”.

Comments from our friends at NZ SMC:

Dr Simon Jackson, Phage Therapy Research Group lead, Waikato University, said:

“The generative AI boom already impacting many aspects of our lives and could one day be used to design viruses that help save lives. Generative AI excels at tackling complex problems that are difficult for humans or conventional computational methods to solve. Biology is full of such challenges and AI is already helping facilitate remarkable progress. A well-known example is the 2024 Nobel prize in Chemistry, which recognised transformational advances in computational protein design and structure prediction.

“The study here goes beyond proteins and asks whether generative AI can design entire functional genomes—the blueprints for life. The focus is a special group of viruses called bacteriophages, or phages, that infect bacteria but are considered harmless to humans. The topic is of particular interest to me, as my group at the University of Waikato is exploring how natural, engineered and synthetic phages might contribute to future treatments for difficult bacterial infections, including antibiotic-resistant superbugs.

“The work is an impressive proof of principle. It combines emerging AI-based genome design ‘software’ with laboratory hardware for synthesising DNA and ‘rebooting’ it into functional phages. Predicting the required combination of thousands of A, C, G, and T DNA bases needed to produce a viable phage is a challenge ideally suited to AI. Even so, only around 5% of the designs worked and half of the functional phages had acquired mutations, suggesting that natural evolution assisted in polishing those AI-generated designs.

“The potential implications for phage therapy are exciting. Most phage therapy development begins by searching nature for suitable phages, and in many cases improving them through laboratory evolution or genetic engineering. Suitable phages can sometimes be exceedingly difficult to find in nature. Generative AI could reduce this dependence on natural discovery and generate useful properties that might be rare or absent in nature.

“Overall, this is an exciting, yet early proof of concept. For practical reasons, the study used a small, well-studied phage and non-pathogenic laboratory bacteria. Designing larger phages, reliably controlling which bacteria they will infect, and proving safety and effectiveness in patients are still to come. There are also important questions around biosafety, public acceptance, and equitable access. Aotearoa New Zealand has traditionally taken a cautious approach to genetic modification, but the Gene Technology Bill currently in progress will introduce a new regulatory system. AI-designed phages are exactly the kind of emerging technology that will test whether reform can support innovation while maintaining public confidence and environmental safeguards.”

Professor Jasna Rakonjac, School of Food Technology and Natural Sciences, Massey University, said:

“Increasing resistance to antibiotics could lead us to a potential dark scenario of untreatable infectious diseases by 2050. Phage therapy is an alternative approach to treating infectious diseases caused by bacteria.

“Bacteriophages or phages are viruses that exclusively attack and kill bacteria, but are completely harmless to higher organisms, from yeast to humans.
 
“One issue with phage therapy is that bacteria also develop resistance to bacteriophages. This can be overcome by various strategies, including applying “cocktails” of multiple phages with the end-purpose to eliminate the bacterium causing the disease. These strategies, however, are often not very effective, leading to inability to treat the infection successfully.
 
“The authors of the Science paper used generative AI to design thousands of variants of a specific bacteriophage species, and tested if some of them will effectively kill a specific strain of E. coli. Among those that were created, a dozen or so indeed were able to kill the intended E. coli strain. Moreover, the authors identified E. coli that became resistant to the new phages and used generative AI to “upgrade” these bacteriophages to overcome bacterial resistance. The AI approach appeared to be much more effective than the usual “cocktail” approach.
 
“Overall this work shows that generative AI is a very powerful tool for creating new variants of bacteriophages as antibacterial therapeutics. Each bacteriophage “species” exists in nature in a myriad of variants, which were sourced here and combined with additional variation using “generative genomics”, an AI-based strategy. Generative AI provides the advantage of sourcing the biological solutions much more effectively than it would be possible under the normal circumstances where potential useful variants of the phages and their genes are geographically or historically too distant to allow natural recombination and evolution to improve their therapeutic potential.”

‘Generative design of bacteriophages with genome language models’ by Samuel H. King et al. was published in Science at 19:00 UK time on Thursday 6 August 2026. 

DOI: 10.1126/science.aec2657

Declared interests

Jordi García Ojalvo: “He declares that he has no conflicts of interest”

Marc Güell:  “He has not said whether he has any conflicts of interest.”

Dr Simon Jackson: “No conflicts of interest to declare.”

Professor Jasna Rakonjac: “No conflicts of interest; I do not work on phage therapy and do not know the authors.”

For all other experts, no reply to our request for DOIs was received.