Typology of Research on Artificial Intelligence Lifecycle Security

Artificial intelligence is vital technology — it offers unparalleled insights but also involves unparalleled security risks.

Typology of Research on Artificial Intelligence Lifecycle Security (TRAILS) is a National Science Foundation-funded project that uses a participatory, lifecycle approach to identify and address evolving security risks in Artificial Intelligence (AI) research.

TRAILS addresses the entire pipeline of AI research —from the collection of training data to the refinement of models — while leveraging cross-disciplinary and cross-sector expertise from academia, government, and industry.

Through the novel integration of qualitative, computational, and participatory methodologies, TRAILS is producing:

1. A validated typology of security risks arising throughout the AI research lifecycle.
2. Training and protocols to help practitioners, policymakers, and government personnel effectively address these risks.

TRAILS's approach includes:

• Cultivating novel sources of firsthand empirical data.
• Strengthening theoretical frameworks for understanding research security.
• Building interdisciplinary communities of practice.
• Contributing to discipline-specific research-on-research-security scholarship.

The methods and frameworks developed are transferable, enabling application at other research institutions and in high-stakes domains.

Get Involved

Fill out the contact form to learn more or participate in this initiative.