# IMDA official introduction: the AI Verify governance testing framework

- Date: 2023-01-15
- Who: IMDA Singapore (Infocomm Media Development Authority (IMDA), Singapore) · IMDA Singapore
- Source: https://www.youtube.com/watch?v=8NfNN1heFQY
- sgai: https://sgai.md/videos/v035/
- License: sgai-authored content (summaries, translations, analysis) is CC BY 4.0 — attribute and link to sgai.md. Verbatim source text (Hansard, speeches, transcripts, policy documents) remains © its original rights holders and is reproduced for reference only. Terms: https://github.com/meltflake/sgai/blob/main/DATA-LICENSE.md

## Why it matters

IMDA launched the world's first AI Verify testing toolkit in 2023, using voluntary self-assessment to replace mandatory legislation, to pre-empt the rule-making authority in international governance standards

## Summary

Official IMDA video introducing the features and applications of AI Verify — the world's first AI governance testing framework and toolkit.

## Key points

- AI Verify is Singapore's international pilot governance testing framework and toolkit.
- The toolkit tests fairness, robustness and explainability, plus process checks on accountability, safety and oversight.
- AI Verify runs in a company's local environment to protect IP and data confidentiality.

## Full text

© IMDA Singapore — reproduced for reference only.

[Music] from virtual assistants on our mobile phones recommendations from your favorite online store to self-driving Vehicles it is clear that artificial intelligence is present in many aspects of Our Lives as more companies use AI in their products and services fostering greater trust in AI is key to building confidence and unlocking transformative opportunities for your business to help companies build trust with their stakeholders Singapore has launched AI verified and AI governance testing framework and toolkit through objective technical tests and process checks AI verify can help demonstrate that your AI applications meet your declared performance benchmarks MDA invites interested companies to participate in this International pilot to test their AI models and share their feedback currently available as a minimum viable product AI verify has two components a governance testing framework and a software toolkit the framework is aligned with internationally accepted AI governance principles AI verify does not define ethical standards instead it allows AI developers users and owners to validate their claims about their AI systems against these principles [Music] the interface guides you step by step through the testing process and is simple enough for both Technical and non-technical users [Music] tests for three principles fairness robustness or explainability the guided workflow makes it easy to configure the test scenarios by helping you choose the most relevant outcomes for your use case using fairness as an example the tests will help assess whether an AI model produces different outcomes across demographic groups such as gender or race beyond the technical tests there is a process checklist for you to self-assess against all the governance principles for instance accountability Safety Human agency and oversight the toolkit provides a summary of the principal outcomes results of the technical tests and Analysis of the process checks to preserve the confidentiality of your IP and data AI verify runs within your local environment how does AI verify work AI verify is very simple to use first you enter the model details and parameters then configure how the toolkit would connect your AI model and upload your test data set this will be used to identify the list of features you can then run the test [Music] using explainability as an example you can indicate whether you have documented the considerations such as rationale risk assessments and trade-offs in selecting a given AI model you can also provide feedback to ensure that the framework continues to be relevant to the industry once technical tests and the process checklists are completed you can view the results and download the report there are a number of benefits for companies who take part in the pilot you'll have early and full access to an internationally aligned AI governance testing framework and toolkit the toolkit will help you conduct self-testing on your AI models and produce reports to help you build trust with your stakeholders you will also enjoy opportunities to network and collaborate with other participating companies when you share your feedback and the reports with imda it will help sheep AI verified and contribute to the development of international standards in AI governance join Singapore's AI verify pilot today to build trusted AI [Music] find out more at go.

gov. sg AI verify