- Home
- AI governance tools
- AI Verify
AI Verify
Singapore
What it is
AI Verify is an open-source toolkit for testing AI-system performance, with more than eight technical tests covering fairness, explainability and robustness. Its framework assesses responsible implementation against 11 AI-governance principles, including transparency, safety, security, data governance and accountability. Users can run tests through a Portal or command line, upload results, complete process checklists and create standard or customized reports. AIVT 2.0 accepts tabular and image data, using a model file, preprocessing pipeline or model API as input. Plugins can extend the toolkit; its Veritas integration adds threshold trade-off tests for fairness and transparency, plus explainability plots. The framework maps to references including NIST AI RMF, the NIST Generative AI Profile, the Hiroshima Process Code of Conduct and ISO/IEC 42001. The Toolkit focuses on traditional machine-learning models; Project Moonshot separately assesses LLM applications through benchmarking and red teaming. AI Verify is available under the Apache 2.0 license. It does not set ethical standards or guarantee that a tested system is safe or free of risks or bias.
For and against
For
- Provides tests for fairness, explainability and robustness.
- Supports tabular and image data.
- Offers Portal and command-line testing.
- Reports can be customized.
- Apache 2.0 license permits open-source use.
Against
- Does not define AI ethical standards.
- Does not guarantee systems are risk-free or unbiased.
- Toolkit tests traditional AI; Moonshot covers LLM applications.
The verdict
AI Verify offers an extensible, open-source way to test traditional AI models and report results against a broader governance framework. Its tests do not establish ethical standards or certify that systems are safe.
Who it suits
The toolkit is built for data-science and compliance users, and supports both self-assessment and third-party testing. AI owners, developers, researchers, service providers and companies integrating it into their systems are also listed as potential users.
On the record
- Data support
- AIVT 2.0 currently supports tabular and image data types and can use a model file, preprocessing pipeline or model API as input.aiverify-foundation.github.io · read 1 Oct 2026
- Extensibility
- Users can add testing algorithms and process checklists to existing Toolkit functions.aiverifyfoundation.sg · read 2 Oct 2026
- Foundation status
- AI Verify Foundation is a not-for-profit wholly owned subsidiary of Singapore's Infocommunications Media Development Authority.aiverifyfoundation.sg · read 2 Oct 2026
- Framework coverage
- The framework assesses responsible implementation against 11 internationally recognised AI-governance principles and covers traditional and generative AI.aiverifyfoundation.sg · read 2 Oct 2026
- Framework principles
- The framework contains 11 AI-ethics principles, including transparency, explainability, safety, security, robustness, fairness, data governance and accountability.aiverifyfoundation.sg · read 1 Oct 2026
- Generative AI tool
- Project Moonshot assesses LLM applications through benchmark testing and red teaming and implements benchmarks recommended in IMDA's Starter Kit.aiverifyfoundation.sg · read 2 Oct 2026
- Integration
- Veritas was integrated into AI Verify to help financial institutions meet common safety-baseline and financial-testing requirements from MAS.aiverifyfoundation.sg · read 2 Oct 2026
- International alignment
- The framework is aligned with internationally recognised principles and frameworks from the EU, OECD and Singapore.aiverifyfoundation.sg · read 1 Oct 2026
- License
- The toolkit is open-sourced under the Apache 2.0 permissive license.aiverifyfoundation.sg · read 1 Oct 2026
- Limitation
- AI Verify does not define AI ethical standards and does not guarantee that tested systems are free from risks or biases or completely safe.github.com · read 2 Oct 2026
- Moonshot interfaces
- Project Moonshot offers a Web UI, interactive CLI, library APIs and Web APIs for MLOps integration.github.com · read 2 Oct 2026
- Open source
- The Toolkit is open-sourced under the permissive Apache 2.0 license.aiverifyfoundation.sg · read 2 Oct 2026
- Portal and CLI
- Users can run tests through the Portal or command line, upload results, complete process checklists and generate reports.aiverify-foundation.github.io · read 1 Oct 2026
- Purpose
- AI Verify is an AI governance testing framework and software toolkit that validates AI-system performance against internationally recognised principles through standardised tests.github.com · read 2 Oct 2026
- Reporting
- AI Verify includes standard report templates that predefine report layouts, technical tests and process checks, while also allowing customized reports.aiverify-foundation.github.io · read 1 Oct 2026
- Reports
- The Toolkit creates customised reports and can combine independent developer and compliance-team work into a full report.aiverifyfoundation.sg · read 2 Oct 2026
- Scope limit
- Project Moonshot tests LLM-based applications, while the AI Verify Testing Toolkit tests traditional AI applications for fairness, explainability and robustness.aiverifyfoundation.sg · read 1 Oct 2026
- Security
- The Foundation says electronic storage and transmission of personal data are secured with appropriate security technologies.aiverifyfoundation.sg · read 1 Oct 2026