About AIRE

Rapid progress in AI has raised concerns about the potential risks associated with its increasing capabilities and widespread deployment. The AI Risk Explorer (AIRE) helps navigate this uncharted territory.

Mission

AIRE provides curated information on the AI risk landscape to enable effective decision-making. To advance this mission, we systematically gather emerging evidence and deliver actionable insights.

Vision

AIRE aims to be the go-to resource for policymakers and experts seeking to better understand AI risk, while also serving as a hub for generating early-warning signs of emerging threats.

Scope

We cover four prominent risks:

Risk Overviews

Conceptualization of each risk through taxonomies, risk thresholds, threat scenarios, and others.

Evaluations

Reports assessing AI models to extract key findings and determine the level of risk across categories.

Benchmarks

Compilation of automated tests and interpretation of model performance on selected ones.

Threats

Real-world incidents and instances of misuse, particularly cyber and influence operations.

Policies

Government and company initiatives dedicated to conceptualizing, assessing, and mitigating each risk.