Just_Super/Getty ImagesThe recent announcement of novel, viable viruses created by artificial intelligence (AI) was celebrated as a major advance in the fight against antibiotic resistance. But it also raised urgent concerns about regulation.Alongside precision gene-editing technologies such as CRISPR/Cas9 and synthetic biology, AI is now making it increasingly possible to design and alter entire biological systems. These technologies could clearly do a lot of good. The new AI-designed viruses – bacteriophages that exclusively kill bacteria – rapidly overcame antibiotic-resistant strains of E. coli.Gene-editing itself is already contributing to medical advances such as the recently approved CRISPR-based therapy for sickle cell disease. And synthetic biology is helping to tackle environmental challenges by employing algorithms to analyse data much faster.But greater capability also brings greater risks of unintended consequences. The same tools may be deliberately misused or unwittingly create risks existing regulations were never designed to anticipate.While the science is advancing rapidly, the rules governing biotechnology are not keeping up, in New Zealand and elsewhere.The challenge is not simply AIIt would be much easier if this was merely an AI safety problem. But there is a more fundamental issue at play: what happens when computational technologies become part of biological design?CRISPR has already changed genetic engineering by allowing scientists to make targeted changes to DNA. Synthetic biology has expanded what scientists can do further, allowing them to design and construct biological systems. AI adds another layer, helping researchers analyse biological information, identify patterns and generate novel genomes.The result is a convergence of technologies that makes the boundary between computational design and biotechnology less clear. This matters for regulation because laws are often organised around particular technologies, organisms or activities. But what happens when a new biological capability emerges from a combination of technologies that regulators have traditionally considered separately?Benefits and risks are connectedThe promise of biotechnology is also what makes it difficult to govern. This is known as the problem of “dual use”: the same technology that helps solve a problem can sometimes create new ones.The research on AI-designed viruses provides a useful example of this dilemma. The researchers were pursuing a potentially useful scientific goal, demonstrating that AI could help design novel bacteriophages and ultimately advance the fight against antibiotic-resistant microbial pathogens. But the technology could be misused and some of the resulting risks may be difficult to predict, detect or contain. The important question is not only what this technology can do now, but what it could enable in the future. For regulators and policy makers, the task is more complicated than deciding whether a technology is “safe” or “dangerous”. They need to find ways to support useful research while identifying and managing risks that can be reasonably anticipated. As biotechnology becomes more capable, finding that balance becomes more important but also more difficult.Balancing innovation and riskExisting rules are not without value. They provide important safeguards for research, manage known risks and allow governments to distinguish between activities that pose different levels of harm. But some of the rules governing biotechnology were designed for a very different scientific landscape.New Zealand is already grappling with this question. The Gene Technology Bill, currently stalled in cabinet because of disagreements between coalition parties, would replace parts of the 30-year-old regulatory framework developed under the Hazardous Substances and New Organisms Act. Rather than treating all gene technologies in the same way, the proposed system would take a more flexible approach. Some lower-risk gene-editing technologies would be exempt from regulation, while activities deemed to pose greater risks would remain subject to controls.This approach illustrates that biotechnology governance and regulation do not have to restrict innovation. New Zealand is not alone in facing this challenge. In Australia, the Gene Technology Act 2000 provides a framework for managing risks from gene technology, with a gene technology regulator overseeing these risks while other agencies regulate specific products, including medicines and agricultural products. But there are regulatory gaps, particularly regarding the editing of the human genome. Similar challenges play out in the United States. Its Coordinated Framework for the Regulation of Biotechnology brings together several agencies, with each retaining its own regulatory role. The biotechnology’s classification determines which agency is responsible, but even a CRISPR-edited mushroom designed to be “brown resistant” fell outside the US Department of Agriculture’s oversight because it did not meet the agency’s definition of a regulated article. The European Union has traditionally taken a particularly cautious approach to biotechnology, with gene technology rules described as “among the strictest in the world”. But since the advent of CRISPR, the EU has adopted new rules, creating simpler pathways for lower-risk technologies while maintaining greater oversight for more complex ones. This shift shows how even highly protective regulatory systems recognise the need to evolve alongside biotechnology. Governance systems should be able to distinguish between different levels of risk, allowing beneficial research to proceed while subjecting high-risk activities to greater scrutiny. The remaining challenge is to ensure these categories remain flexible enough to accommodate technologies as they continue to evolve. The goal, therefore, is not to choose between innovation and regulation, but to develop governance that can recognise emerging risks without sacrificing the potential benefits biotechnology has to offer.Chelsea R. Francek does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.