AI Verify Foundation

AI Verify Foundation (AIVF) is a not-for-profit entity that works to “build trust through ethical AI” by convening and mobilising a global open-source community to create AI testing tools and advance responsible AI practice. The Foundation’s stated aim is to “harness the collective power and contributions of an international open-source community to develop Artificial Intelligence ("AI") testing tools to enable the development and deployment of trustworthy AI,” and it also “promotes best practices and standards for AI.” AIVF describes itself as “a global open-source community that convenes AI owners, solution providers, users, and policymakers, to build trustworthy AI.” Structurally, AIVF is “a wholly owned subsidiary of the Infocommunications Media Development Authority of Singapore (IMDA).”

AIVF’s mission is organised around community-building and shared infrastructure for AI testing and governance. It seeks to “Foster a community to contribute to the use and development of AI testing frameworks, code base, standards, and best practices,” while also creating “a neutral platform for open collaboration and idea-sharing on testing and governing AI.” In addition, it aims to “Nurture a network of advocates for AI and drive broad adoption of AI testing through education, outreach, and marketing.” The Foundation also formalises membership and governance: organisations can join as “Premier Member or a General Member,” and the “Governing Committee” includes “One representative appointed by each Premier Member” and provides guidance on AIVF’s strategic direction and product roadmap.

On the tooling side, AIVF points to AI Verify as “An AI governance testing framework and software toolkit that help industries be more transparent about their AI to build trust,” positioning it as practical infrastructure for transparent and trustworthy AI development and deployment. For generative AI specifically, AIVF operates Project Moonshot, described as “one of the world’s first Large Language Model (LLM) Evaluation Toolkits.” Project Moonshot is “an open-source tool to bring benchmarking and red teaming together to help LLM App developers and compliance teams to test and evaluate their LLMs and LLM applications,” and is hosted on GitHub. The project supports implementation of “IMDA’s Starter Kit,” provides “more than 100 (and still growing) benchmark datasets,” and is designed to help organisations answer questions about which foundation model to use and how to ensure an application is “robust and safe,” producing “benchmarking tests and scoring reports” so that developers and compliance teams can deploy “responsible Generative AI systems with confidence.” The Moonshot page also explicitly links the effort to AI safety practice, characterising the project as “one of the first tangible representations globally of what it means to approach AI Safety, in a way that is actionable for companies and AI teams.”

Beyond open-source tools, AIVF also runs initiatives intended to develop practical norms for technical assurance of real-world generative AI applications. Its Global AI Assurance Pilot is “a global initiative launched in February 2025 to help codify emerging norms and best practices around technical testing of Generative AI applications.” The pilot’s design includes pairing “AI assurance and testing providers with organisations deploying Generative AI applications,” focusing on “technical testing of the real-life application (not the underlying foundation model),” and using “lessons learnt from specific examples to create generalisable insights on ‘what and how to test’.” The pilot explicitly enumerates risk dimensions to consider during testing (e.g., “Inappropriate Data disclosure,” “Malicious use,” “Other security risks,” and “(Lack of) Appropriate level of human oversight”), and frames “Targeted outcomes” such as “Inputs into future standards for technical testing of Generative AI applications” and “Inputs into open-source and proprietary testing tools for Generative AI applications.”

AIVF’s membership charter further sets out how it manages open-source contributions and sharing. For example, “All new inbound code contributions to AIVF shall be made under the Apache License, Version 2.0,” “All outbound code will be made available under the Apache License, Version 2.0,” and “All technical documentation pertaining to the AI Verify Toolkit will be contributed to and made available by AIVF under the Creative Commons Attribution 4.0 International License.” Together, these elements position AIVF as an organiser of a multi-stakeholder, open-source ecosystem for AI testing frameworks, LLM evaluation and red teaming, and the codification of emerging testing norms and standards intended to enable trustworthy and responsible AI deployment.