AIxBio = Big Promise + New Risks

Artificial intelligence (AI) is transforming the life sciences and public health: accelerating drug discovery, vaccine development, and outbreak detection. But with rapidly advancing capabilities comes unprecedented risks — these same tools could be misused to design dangerous pathogens.

NTI’s tabletop exercise at the Munich Security Conference this past February explored risks and opportunities at the convergence of AI and the life sciences. The scenario: a fictional extremist group uses AI-enabled tools to engineer a novel virus that triggers a global pandemic, resulting in more than 850 million cases and 60 million deaths.

The takeaway: the scenario represents a plausible near-term risk that technical experts and counterterrorism professionals find deeply concerning. It’s a risk that the world isn’t ready for.

Here’s what needs to happen:

1. Fund a Global R&D Agenda to Develop AIxBio Guardrails

National governments, industry, and philanthropy should invest in research to build technical and policy guardrails for AI-biology tools.

• Develop built-in screening mechanisms and managed-access frameworks for biological design tools.
• Pair technical solutions with innovative governance measures, including incentives and accountability frameworks.
• Support horizon scanning to track emerging risks and adapt safeguards as technology evolves.

Innovation and security must advance together – and that requires innovation in safety and security. 

2. Create National Institutes for AI Safety and Biosecurity

Countries with AI investments should establish institutes focused on AI safety and security, with biosecurity as a core mission.

• Build on commitments from the 2024 AI Seoul Summit.
• Integrate biosecurity expertise with AI safety competencies.
• Act as hubs for research, public-private collaboration, and development of novel governance approaches.

These institutes should also serve as independent evaluators of industry-proposed safeguards, ensuring they remain responsive to the evolving threat environment.

3. Launch a Global Platform to Disseminate Best Practices for AIxBio safeguards

AI-enabled biology is a global challenge that demands a global response.

• Create an international forum to share norms, tools, and standards for responsible AIxBio innovation.
• Engage industry, academia, and policymakers to build trust and encourage the adoption of best practices.
• Draw lessons from the DNA synthesis industry’s involvement in shaping proactive governance approaches that balance innovation with security.

NTI’s AIxBio Global Forum is a great starting point. Scaling it up could help industry and policymakers globally prevent misuse of AIxBio capabilities while preserving their benefits.

4. Use AI to Strengthen Biosecurity and Pandemic Preparedness

AI isn’t just a risk; it’s a powerful defense tool.

• Deploy AI for biosurveillance to detect outbreaks early, even engineered ones.
• Accelerate development of vaccines and therapeutics to mitigate biological threats.
• Use AI-driven analytics to identify and disrupt malicious actors before they strike.

Using AI for defense can make bioweapons less appealing—and help ensure the next pandemic is stopped before it starts.

The Bottom Line

The AI-biology convergence offers enormous benefits but also brings about risks as we’ve never seen before. Without action from multiple disciplines, the race for AI development and dominance could become a race to the bottom when it comes to safety and security.

The time for guardrails is now.