The whack-a-mole governance challenge for AI-enabled synthetic biology: literature review and emerging frameworks

Trond Arne UndheimarXiv 2024 · DOI · arXiv

Introduction

Synthetic biology (synbio) stands at a critical juncture. Long confined to niche laboratory experiments, the field is rapidly transitioning toward industrial-scale biomanufacturing. The primary catalyst for this shift is the convergence of synbio with artificial intelligence (AI), particularly generative models. While AI promises to solve the scaling bottlenecks that have historically limited the bioeconomy, it simultaneously introduces a "whack-a-mole" governance challenge. As soon as one biorisk is mitigated, the accessibility and speed of AI-driven design allow new threats to emerge. This paper explores the necessity of moving from traditional "command-and-control" oversight toward a distributed, adaptive stewardship model.

Methodology

To map the current landscape of AI-enabled synbio governance, a systematic literature review was conducted following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines.

PRISMA flow diagram illustrating the systematic literature search and selection process.

The search spanned multiple databases including Google Scholar, PubMed, and Scopus, focusing on literature published between 2020 and 2024. From an initial pool of 653 abstracts, 204 relevant sources—comprising peer-reviewed papers and high-impact "grey literature"—were selected for qualitative synthesis. The analysis involved extracting over 1,200 unique keywords, which were subsequently categorized into thematic clusters to identify the primary tensions in the field.

The Convergence of AI and Synthetic Biology

The integration of AI into biological workflows has fundamentally altered the "lab" environment. We are witnessing the rise of the dispersed lab, where biological design no longer requires physical proximity to high-end equipment. Advanced insights that previously necessitated doctoral-level expertise are now accessible via automated design tools and large language models (LLMs).

This shift renders traditional governance regimes, such as the Dual Use Research of Concern (DURC) framework, increasingly obsolete. These frameworks were designed for a pre-AI era characterized by centralized research and slow publication cycles. In contrast, the current reality involves "wetlabs and foundries on demand," where the barrier to entry for potentially hazardous research is significantly lowered.

Key Findings and Issue Clusters

The literature review identified eight core "issue clusters" that define the current state of synthetic biology. These clusters represent the various dimensions that governance must address, ranging from the technical tools and scientific foundations to the broader socio-economic impacts on the bioeconomy and national security.

The eight thematic clusters identified in the literature review, ranging from Tools and Risks to Governance and Bioeconomy.

A critical finding is the significant gap in policy research. Despite the ubiquity of AI in biological practice, only a small fraction of the sampled papers explicitly address the specific risks posed by generative AI and LLMs. Furthermore, the "business case" for synbio remains under-theorized, with a conspicuous absence of management literature regarding the commercialization of these technologies.

Proposed Governance Framework: The 6x4 Model

To address the "whack-a-mole" nature of emerging biorisks, this paper proposes a 6x4 Governance Framework. This model maps six levels of actors against four distinct governance types:

  1. Levels of Engagement: Global, National, Corporate, Lab/Startup, Scientist/Network, and Citizen.
  2. Governance Types:
    • Precautionary: Hard law and formal regulations.
  3. Stewardship: Soft law, norms, and professional ethics.
  4. Bottom-up: Adaptive, community-led responses.
  5. Laissez-faire: Industry-driven standards and market forces.

The goal is to move toward "active stewardship," where responsibility is shared across the entire ecosystem rather than being solely the domain of government regulators.

Conclusion and Future Directions

The transition to a "gigascale" bioeconomy requires more than just biological and computational breakthroughs; it requires a fundamental rethink of oversight. Current limitations include a "data gap," where AI development is hindered by proprietary or poorly formatted biological datasets. Future efforts must focus on creating "AI-ready" open-source data and fostering a culture of distributed vigilance. Governance in the age of AI-enabled synbio is not a problem to be "solved" once, but a continuous process of adaptation to keep pace with the rapid evolution of the technology.

Summary updated

BibTeX
@misc{undheim2024whack,
  author = {Trond Arne Undheim},
  title = {{The whack-a-mole governance challenge for AI-enabled synthetic biology: literature review and emerging frameworks}},
  year = {2024},
  eprint = {2402.00312},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2402.00312},
}
Figure