Open Source

Flaw Reporting for AI
FLARE-AI

We are researchers trying to streamline AI flaw and incident reporting, to connect issues with those who can fix them.

Partners & Collaborators

Our solution

We teamed up with AI developers, AI researchers and red teamers, software security professionals, and civil society organizations to not only flag AI safety issues, but also contribute necessary information for developers to effectively triage the problem. Our flaw reporting tool lets anyone create and download their own flaw report, or submit to a range of stakeholders.

The people behind it

Team

Lead Authors

Shayne Longpre

MIT

Elaine Zhu

Northeastern University

Carson Ezell

Harvard University

Avijit Ghosh

Hugging Face

Top Contributors

Sean McGregor

AVERI

Kevin Paeth

UL Research Institutes

Kevin Klyman

Stanford / Harvard University

Sayash Kapoor

Princeton University

Rishi Bommasani

Stanford University

Ruth E. Appel

Stanford University

Contributors

Gregory Strom

Carnegie Mellon University SEI CERT

Lauren McIlvenny

Carnegie Mellon University SEI CERT

Mark M. Jaycox

Google

Peter Slattery

MIT FutureTech

Nathan Butters

ARVA

Advisors

Arvind Naryanan

Princeton University

Percy Liang

Stanford University

Alex Pentland

MIT

What is AI Flaw Reporting?

AI flaw reporting is the process of identifying, documenting, and disclosing safety issues, security vulnerabilities, or harmful behaviors in artificial intelligence systems. This includes issues like:

Security vulnerabilities that could be exploited by malicious actors
Safety issues that could cause real-world harm
Bias, discrimination, or unfair treatment in AI outputs
Unintended behaviors or capability overhang
Privacy violations or data mishandling

Purpose of This Website

This platform provides a structured, standardized way to document AI flaws and incidents. Our goal is to:

Create comprehensive reports that can be submitted to appropriate organizations
Ensure consistent documentation standards across the AI safety community
Facilitate responsible disclosure to AI developers and safety organizations
Build a knowledge base for improving AI safety practices

Collaborators & Integrations

Who we are and why we're doing this

Researchers from MIT, Stanford, Harvard, Northeastern, and Hugging Face built this tool, in collaboration with cybersecurity experts, to allow the community to submit fully standardized AI flaw reports to stakeholders across the ecosystem.