Research
The latest research from Americans for Responsible Innovation.
The Invisible Backbone
AI has increasingly become an underlying infrastructure for the United States, yet the AI sector lacks formal recognition as critical infrastructure, leaving it without public-private coordination mechanisms, information sharing, asset mapping, and protection standards commensurate with AI’s increasing role in critical operations.
Responsible Innovation at the Frontier
ARI’s blueprint for federal AI governance is designed to promote safe frontier AI development in America. The blueprint is built around three governance functions any federal proposal should incorporate.
PDF
Information is Power: The Case for Data Center Transparency
Load growth from computational demand is outpacing the ability of grid planners, energy analysts, and
policymakers to track, model, and manage its consequences.
PDF
Growing Up With Chatbots: Why We Need National Minor Safety Benchmarks
The same features that
make AI chatbots appealing: around-the-clock availability, non-judgmental responses, and the ability to
mimic emotional understanding, also heighten risks for young users who are still developing critical thinking
skills, emotional regulation, and healthy relationship patterns.
PDF
Hands on the Trigger: The Operational Necessity of Humans with Lethal Autonomy
This report finds that meaningful human control over autonomous weapon systems is a necessary, operationally achievable safeguard that preserves command accountability, reduces the risk of unrecoverable failure, and can be maintained without sacrificing tactical speed or strategic advantage.
PDF
Competitive Compliance: Why Uniform Screening Standards Support Innovation and Thwart Regulatory Capture
This report finds that mandatory gene synthesis screening is a necessary, cost-effective biosecurity safeguard that closes a critical federal gap, strengthens market integrity, and can be implemented with minimal burden or risk of regulatory capture.
PDF
Proactively Developing & Assisting the Workforce in the Age of AI
Research examines areas of workforce and AI policy that lawmakers should explore, including data and measurement, workforce development and education, improved social safety nets, and place-based and industry-level policies.
PDF
The Stick, the Carrot, and the Net: Policy Approaches for Addressing AI Agent Harms
This report examines three policy approaches for addressing AI agent harms: the “stick” of conventional tort liability; the “carrot” of liability immunity in exchange for proactive governance measures; and the “net” of no-fault compensation schemes that provide swift remedy without fault attribution.
PDF
Recommendations to OSTP for Gene Synthesis Screening Framework
This report examines pathways for implementation of the Trump Administration's May 5th Executive Order, Improving the Safety and Security of Biological Research, offering policy ideas that make it more difficult for bad actors both domestically and globally to access genetic material that would allow them to produce deadly pathogens.
PDF
State-Level AI Laws May Drive Adoption of AI Tools
State-level AI legislation does not hinder public interest in and adoption of generative AI tools—in fact, it may help drive it. As policymakers weigh the risks and benefits of AI regulation, this paper analyzes AI search data and AI regulation.
PDF
Federal Preemption of State Laws
Author Iskandar Haykel examines how state laws would be impacted by federal legislation that would preempt state AI regulation for the next 10 years. This research explores case studies of state tech laws in areas that include consumer protections, developer transparency, kids online safety, and more.
PDF
Assessing AI’s Labor Disruption
This report explores the gap between economist forecasts of AI's labor disruption and technologists' predictions, finding that the difference can be partially explained by failures of the General Purpose Technology framework in assess AI's impact.
PDF
Transparency in Frontier AI
ARI ranks seven big-name AI models on measures of transparency. The study explores transparency in four primary areas, including user-facing documentation, risk and safety, technical transparency, and evaluation and impact. The research uses 21 metrics to rank models in each of the four areas.
