AI Risk and Vulnerability Alliance (ARVA)
General inquiries: Munyer@airava.org
● A nonprofit for safer, accountable AI

Building the shared knowledge needed to make AI systems safer.

The AI Risk and Vulnerability Alliance (ARVA) is a nonprofit organization working to empower communities to recognize, diagnose, and manage vulnerabilities in AI systems that affect them.

We work across research, technology, policy, and civil society to advance practical AI risk management.

Open knowledge — Shared resources for AI risk and vulnerability
Cross-disciplinary — Research, practice, policy, and community
Public interest — Transparency, accountability, and safety
Our mission

Empowering communities to address vulnerabilities in AI.

ARVA works to make the risks and failures of AI systems visible, understandable, and actionable. We believe safer AI requires shared knowledge, independent oversight, and meaningful participation from affected communities.

01

Recognize AI risks

We help surface harmful flaws, security weaknesses, and systemic vulnerabilities across AI models, datasets, systems, and deployment practices.

02

Diagnose failures

We develop shared taxonomies, research, and knowledge resources that make AI risks easier to understand, document, and evaluate.

03

Manage vulnerabilities

We support practical approaches to AI risk management, including evaluation, auditing, disclosure, mitigation, and responsible governance.

Our work

Turning AI risk into actionable knowledge.

ARVA connects technical research with the needs of practitioners, policymakers, and communities affected by AI systems.

How we advance AI safety

  • ✓ Conduct and support research on AI risks, vulnerabilities, and safeguards
  • ✓ Maintain open knowledge resources for AI risk management
  • ✓ Advance evaluation, auditing, and red-teaming practices
  • ✓ Engage researchers, developers, policymakers, and civil society
  • ✓ Promote transparency, accountability, and public-interest oversight
Initiatives

Open resources for AI risk management.

Our initiatives are designed to make AI vulnerabilities more visible and to provide practical tools for safer development and deployment.

Flagship initiative

AI Vulnerability Database (AVID)

AVID is an open-source knowledge base that brings together AI risks, vulnerabilities, and mitigation practices. Inspired by established cybersecurity practices, it provides a structured way to document, understand, and learn from failures in AI systems.

Research & standards

AI vulnerability research

ARVA conducts and supports research on AI vulnerabilities, evaluation, auditing, disclosure, and risk governance. Our work focuses on turning emerging safety challenges into practical, shareable knowledge.

Community & policy

Public engagement and accountability

We engage in public consultations, policy discussions, and community initiatives to advance accountability, transparency, and safer AI systems for the public interest.

Organization

A nonprofit built for public-interest AI safety.

ARVA is a nonprofit organization committed to independent research, open resources, and collaboration across disciplines and communities.

Independent research

We pursue rigorous, practical research on AI risks, vulnerabilities, and mitigation strategies.

Open resources

We create and maintain publicly accessible resources that support shared learning and responsible practice.

Community participation

We engage affected communities and diverse stakeholders in understanding and addressing AI harms.

Responsible governance

We support transparency, accountability, and effective oversight across the AI ecosystem.

Contact

Get in touch with ARVA

For general inquiries, collaboration opportunities, research questions, media requests, or partnership discussions, please contact the AI Risk and Vulnerability Alliance.