Blog18th December 2025
It’s hard to believe more than a year has passed since we published our 2024 horizon scan. We have been busy since then; participating in over 30 events, delivering 6 workshops and reading countless papers and reports to keep our knowledge up-to-date.
And as we approach the end of 2025, I wanted to take a step back and reflect on three predominant themes that have come up in our scanning work this year:
AI in engineering biology
AI is a topic of excitement, trepidation and uncertainty across science. Advancing technological capabilities, in combination with ever-growing datasets, mean AI-driven tools and methods are increasingly being integrated into parts of the research and development process with the aim of accelerating discovery and innovation.
We are now seeing examples where specialised AI tools can be trained on large biological datasets to make predictions and generate new data: earlier this year, a team from MIT reported using generative AI, trained on a dataset of known compounds and antimicrobial activity, to design two new antibiotics for combatting hard-to-treat infections. It is currently unknown whether these two drugs will prove safe and effective, and more broadly, whether AI will truly transform costly and time-intensive drug development pipelines. However, applications like this highlight the potential for AI to be used in this way.
Other research is exploring the application of generative AI to design whole synthetic genomes, chromosomes and living systems. This year, researchers reported creating the world’s first AI-generated viruses and a Wellcome-funded project was launched to develop the tools needed to synthesise human genomes. Both signal initial steps in an emerging field of AI-driven genome design.
The UK Government’s AI for Science strategy, published last month, describes AI models as “increasingly autonomous participants in the scientific process”, able to draft abstracts, propose experiment protocols, conduct analysis, and, with the help of robotics, carry out experiments. This shift in how science is conducted raises fundamental questions about what AI automation means for scientific creativity, methodology, novelty and quality.
A recent OECD report on Synthetic Biology, automation and AI highlights the need for systems to balance automation with effective human intervention to ensure transparency, responsibility, trust and ethical consideration. ‘Humans-in-the-loop’ hold vital roles in understanding and justifying algorithmic decisions, ensuring datasets used are representative so that models benefit everyone equitably, and directing research aims, with consideration of broader societal goals, challenges and values.
Taken together, this shows the convergence of AI and engineering biology continues to raise complex ethical questions as well as unique opportunities, making it a prominent topic in our scanning.
Microbiome science
Researchers continue to uncover the ways in the which the microbes living on, in and around us shape human and environmental health. This year, the Microbiome Innovation Network was launched to strengthen microbiome research and applications across the UK.
Metagenomic analysis – allows scientists to identify both known and unknown or novel microorganisms, and to better understand what they do – revealing information about their evolution and the function of their genes. This represents an important step towards better understanding microbial diversity and translation into clinical applications. For example, researchers are using metagenomics to ‘mine’ the microbiome for new antibiotics, or to understand how we might engineer the human microbiome to protect against infection.
This summer, we ran a Horizon Scanning Lab with the Federation of European Microbiology Societies (FEMS), engaging over 50 scientists and interdisciplinary experts to explore future scientific, ethical and societal implications of microbiology. The group emphasised the importance of research beyond the human gut microbiome, encouraging studies focussed on the underexplored skin, oral and environmental microbiomes, and their interaction with health. The workshop also highlighted the need to develop microbiome stewardship initiatives (e.g. Microbiota Vault, Global Microbiome Conservancy) to conserve microbial resources for equitable research. Many experts across the UK have also suggested a ‘UK Microbiome Biobank’, modelled on the existing and hugely successful UK Biobank, to facilitate microbiome research globally.
Biosecurity
Current trends suggest biosecurity will continue to accelerate as a challenge with new risks emerging. Advances in AI and biotechnology, rising geopolitical tensions, deteriorating trust, and a changing climate all have implications for our resilience towards biological threats. This was spotlighted by Covid-19 and more recently, calls from leading biologists to address the unprecedented risks of ‘mirror life’ microbes.
Frontier AI tools could pose new risks if exploited to synthesise harmful toxins or pathogens. Earlier this year, researchers warned that current screening practices used by DNA synthesis companies to avoid selling materials that could be used to make dangerous proteins are becoming increasingly inadequate. Although screening tools were able to spot natural pathogens and known toxins, novel and potentially dangerous proteins designed by AI could slip through the net.
Other advances in technology can help to mitigate biosecurity risks, particularly around pandemic preparedness and infectious diseases. The cost of metagenomic sequencing has fallen 3 million-fold over the past 25 years, making it viable to deploy at scale for bio surveillance. For example, this year, the UK launched the world’s first pathogen-agnostic metagenomic surveillance system, providing real time data to enable rapid detection of biological and pandemic threats.
Bioengineering approaches are also being explored to prevent the spread of zoonotic diseases. Although not currently permitted in the UK, genome editing can be used on commercial livestock to make them more resilient and less likely to transmit diseases such as bird flu. Gene drives – genetic engineering of disease vectors like mosquitoes to reduce their reproduction and spread – are also being developed. Looking further into the future, researchers are exploring self-disseminating vaccines that could be distributed among wild animal populations. Each of these methods carry complex ethical considerations, for example, around animal welfare, ecological risks, justice and consent.
Our next NCOB timeframed scan will publish in early 2027 and we are excited to be planning the workshops and engagement activities that will feed into it. If you have any thoughts, insights or suggestions for emerging issues we should be aware of, please get in touch via futures@nuffieldbioethics.org.