Saidot's statement of commitment to the AI Pact

By Meeri Haataja (CEO and Co-Founder of Saidot), Brussels, 25 September 2024.

Thank you for the opportunity to share our perspective on the AI Act and its essential role in addressing AI risks.

With AI’s great potential come also major risks. Like that extra finger in an AI-generated image, they’re something that often slip by us. It is all of us in the AI industry who must take accountability: to identify and mitigate those risks, whether ethical, regulatory, or business-related. The AI Act’s risk-based approach is what we need.

So yes, you can count on us to participate in the AI Pact: Having an AI governance strategy to advance the use of AI in compliance with the AI Act, maintaining an inventory of AI use cases, not only high-risk but all of them, and ensuring our team is AI literate. Furthermore, we are committed to helping other organisations to do the same efficiently, ensuring that this necessary process doesn’t delay AI time-to-market for European businesses.

Today I’m excited to share how this work is already proving its value. We have witnessed how AI products’ market readiness can be accelerated with the right kind of governance, on the one hand through better clarity and efficiency of risk management and compliance, and on the other hand by helping develop higher-quality products that are more market-ready. Some of the learnings we have found to be critical are:

• The AI Governance Framework plays a crucial role in defining how and by whom AI governance is implemented within the company;

• We have accelerated AI risk management with a curated AI Risk Library, which collects key information on AI risks, mitigations, and real-world incidents. These risks are not only a matter of safety but also of quality and understanding and mitigating them early on helps to speed up product readiness;

• It is essential to connect AI risks to Model Evaluations, establish quantified methods for measuring model capabilities and safety, and automate these evaluations;

and finally

• We have made AI Act requirements more accessible and understandable for AI teams by creating tailored AI Act Templates that include only the relevant requirements and guidance, rather than all possible ones.

We’ve seen how this process can be streamlined by empowering teamwork and multi-disciplinary collaboration between AI teams, risk, compliance, and auditors – the three lines of defence – and directing governance efforts toward the right things at the right time.

The right thing is also not to govern to avoid all risks but to keep risks in control.

Just like the extra finger in AI-generated images, we need to uncover and address AI’s risks effectively. By doing so, we create well-being while protecting what truly matters: our safety, our rights, and the trust we place in AI and each other.

Saidot's statement of commitment to the AI Pact

More Insights

Saidot launches Integrated Evaluations for Microsoft Azure AI Foundry

Why AI governance is good for the bottom line

Miltton and Saidot enter strategic partnership on responsible AI

Get started with responsible AI.