Panoptic dual-use management: preventing deliberate pandemics in an age of synthetic biology and artificial intelligence

Thesis

Powerful new technologies can have profound global security implications. In this thesis, I investigate how advances in synthetic biology and artificial intelligence could have dual-use potential and enable the deliberate release of pandemic pathogens. I review risks from synthetic biology based on case studies on wildlife virus discovery, viral engineering for vaccine design, and viral engineering for gene therapy. For assessing impacts of artificial intelligence, I consider large language models and biodesign tools. I find that related advances can create new methods to engineer pathogens and make such capabilities increasingly accessible to non-specialists.

These risks are not well captured by existing risk mitigation measures. I argue that the management of dual-use virological research is currently defined by oversight of individual research projects. This is effective for addressing high-risk research but fails to address risks from a more diffuse set of research and technologies with dual-use potential.

To help mitigate these risks, I introduce the idea of panoptic dual-use management. Inspired by methodologies to reduce carbon emissions, panoptic dual-use management involves treating associated dual-use risks as negative externalities and creating appropriate incentives so they are accounted for in decisions between projects. I explore ways in which such incentives could be created for various stakeholders. For instance, funding bodies could use dual-use risks as a tiebreaker between projects on the brink of getting funded, a practice which would incentivise researchers to preferentially propose projects with lower dual-use risks. To realise this proposal, I sketch out a framework for assigning tiered dual-use scores to virological research.

I conclude by highlighting the importance of combining different dual-use management approaches across stakeholders and geographies to establish an effective complex of overlapping mitigation regimes.

Authors: Sandbrink, J

• DOI: 10.5287/ora-bpqk58ead
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: technology governance; pandemic prevention; artificial intelligence; pandemic virus; dual-use risks; biosecurity; dual-use dilemma
• Subjects: Science and technology studies; Virology; Technology governance; Medical ethics; Biosecurity; Medical sciences