October 2, 2025

expert reaction to study on strengthening biosecurity screening for genes that encode proteins of concern

A study published in Science looks at screening for genes that encode proteins of concern to strengthen biosecurity. 

Prof Natalio Krasnogor, Professor of Computing Science and Synthetic Biology, Newcastle University, said:

“I do think this is a risk we must take seriously; the paper clearly makes the case and provides a quite honest assessment of the strength and limitations of their approach. If one moves beyond the more grandiose claims about generative AI these AI protein design tools do pose a growing biosecurity risk because they have the potential to produce functionally dangerous proteins with little homology to sequences of concern. Because current industry screening (the paper calls it Biosecurity Screening Software) is largely homology-centric using a best-match approach, protein sequences that are designed by AI protein design tools and which lack obvious sequence similarity can slip past their screening (as the paper shows). 

“Moving from an AI generated sequence to a high enough yield of a working functional toxin is, however, a far from simple process that requires expertise, equipment, time and money. This is an important limitation of the paper, which the authors acknowledge, because “constructability” of the synthetic proteins has not been considered. It is possible that many of the perceived risky generative proteins of concern are, in practice, not feasible or too difficult to synthesise in sufficient quantities.

“All combined, I think means we do need as a society take this seriously now before additional advances in AI makes the validation and experimental production of viable synthetic toxins also much easier and cheaper to deploy than it is.”

Prof Francesco Aprile, Associate Professor in Biological Chemistry at Imperial College London’s Molecular Sciences Research Hub, said:  

“The study focuses on a pressing issue in protein engineering and biosafety. As AI-driven protein design accelerates, so do the challenges for biosecurity, including the possibility that variants of concern slip past DNA synthesis screening. 

“By introducing targeted improvements to existing software, the authors significantly enhance detection and flagging. 

“This work provides a practical, timely safeguard that strengthens current DNA synthesis screening, and establishes a solid foundation for continued optimisation.”

Prof Daniel McCluskey, Professor of Aerosol Bio-detection Technology at the University of Hertfordshire said:

“1 – This is something to be aware of, but not to be alarmed about. The article has highlighted the issue and, importantly, set out clear approaches for managing it. The fact that the research community has already documented this means we are well placed to monitor and respond effectively.

“2 – The publication outlines practical steps for mitigation, which provide a solid foundation for action. Drawing parallels with the approaches well established in the realm of cybersecurity is an excellent starting point. By following the evidence-based recommendations already available, we can minimise risks and ensure the situation remains under control. What matters is that this has been recognised early and though clearly this requires continued observation and attention from research and biosecurity communities.

“3 – The authors of the study highlighted the fact that nucleic acid synthesis companies already employ biosecurity screening software to screen for known proteins of concern. A significant future challenge lies in developing the capability to predict potential future synthetically created proteins of concern, which could be developed inadvertently, or malevolently, using emerging techniques. “

‘Strengthening nucleic acid biosecurity screening against generative protein design tools’ by Bruce J. Wittmann et al. was published in Science at 19:00 UK time on Thursday 2^nd October.

DOI: 10.1126/science.adu8578

Declared interests

Prof Natalio Krasnogor: besides being a Prof. of Computing and Synthetic Biology at Newcastle University I am also CTO of GitLife Biotech Ltd.

Prof Daniel McCluskey: There are no conflicts of interest.

Prof Francesco Aprile: As part of my academic research, I routinely order DNA from CROs and develop tools for rational protein design.