Regulating genome language models: navigating policy challenges at the intersection of AI and genetics

Area of Science:

• Genomics
• Artificial Intelligence
• Bioethics
• Regulatory Science

Background:

• Genome Language Models (GLMs) integrate AI and genomics, enabling advancements in biological discovery and healthcare.
• Existing AI governance and genomic privacy frameworks are insufficient to address the unique challenges posed by GLMs.
• GLMs create regulatory gaps due to conflicting principles of data access and stringent privacy requirements.

Purpose of the Study:

• To examine the regulatory challenges at the intersection of AI and genomics posed by GLMs.
• To analyze how GLMs impact disease risk prediction, international research, and open-source model distribution.
• To propose a multilayered governance framework for responsible GLM development and deployment.

Main Methods:

• Analysis of regulatory gaps between AI principles and genomic governance.
• Examination of GLM applications in disease prediction, research collaboration, and model distribution.
• Proposal of a hybrid governance framework combining policy and technical solutions.

Main Results:

• GLMs highlight tensions between AI's data access needs and genomics' privacy demands.
• Conventional regulatory approaches are inadequate for GLM applications.
• A multilayered framework integrating policy (e.g., regulatory sandboxes) and technical solutions (e.g., privacy preservation) is proposed.

Conclusions:

• Adaptive governance strategies are crucial for responsible GLM innovation.
• The proposed framework aims to balance innovation with safeguarding individual rights and addressing biosecurity concerns.
• Bridging AI and genomic regulation is essential for navigating the evolving landscape of GLMs.