What the Microsoft outage reveals about AI governance, digital dependency, and the future of trustworthy AI.
The recent Microsoft outage left thousands of people struggling to access essential services, including Outlook and other Microsoft 365 applications. For many organizations, what should have been a normal working day became a reminder of how dependent we have become on technology.
But as I followed the disruption, one question kept coming to mind: If we cannot always rely on the infrastructure supporting our digital systems, what does that mean for the future of trustworthy AI?
We spend considerable time discussing responsible AI, focusing on fairness, transparency, accountability, privacy, and the ethical use of technology. These conversations are necessary, but I believe we sometimes overlook something equally important: the reliability of the infrastructure that makes AI possible.
An AI system may be designed responsibly, tested for bias, and governed by established policies. Yet when the cloud services or external platforms supporting it become unavailable, even a well-governed system may be unable to function as intended. Trustworthy AI is not just about how a system makes decisions. It is also about whether the organization can depend on that system when it matters most.
The Hidden Risk Behind Our Growing Dependence on AI
Organizations are adopting AI at an extraordinary pace. Businesses are automating workflows, universities are integrating AI into teaching and research, and governments are exploring AI-powered services to improve efficiency. These developments are exciting, but they also introduce dependencies that deserve greater attention.
Consider universities, for example. As an AI educator and researcher, I see tremendous opportunities for artificial intelligence to transform learning, research, and institutional operations. However, imagine a university that relies heavily on AI-powered student support, cloud-based learning platforms, and automated administrative processes.
What happens when the infrastructure supporting these services suddenly becomes unavailable?
The same question applies to hospitals, financial institutions, government agencies, and businesses.
The Microsoft disruption did not mean that AI itself had failed. Rather, it highlighted how interconnected digital services have become and why organizations must understand the risks associated with relying on external technology providers. This is where I believe our approach to AI governance needs to evolve.
AI Governance Must Prepare for More Than Ethical Risks
AI governance cannot stop at developing policies, establishing oversight committees, or demonstrating regulatory compliance. Organizations must also consider how their AI systems will respond to infrastructure failures, service interruptions, and unexpected disruptions.
As we move toward more autonomous AI systems, this becomes even more important. An AI agent may be capable of completing complex tasks, but what happens when the applications or services it depends on stop responding? Who takes responsibility for unfinished processes, interrupted decisions, or services that cannot be delivered? These are not simply technical questions. They are governance questions.
Through my work in responsible AI and the development of SAFER AI Protocol, I continue to explore how organizations can translate AI governance principles into practical measures that support accountability, reliability, and responsible adoption.
The recent outage reinforces something I consider fundamental: Trust should not be measured only by how well technology performs when everything is working. It should also be measured by how prepared we are when things go wrong.
This does not mean organizations should slow down AI adoption or abandon cloud technologies. Innovation will always involve some level of risk. The responsibility lies in recognizing those risks and preparing for them. As AI becomes increasingly embedded in our institutions and everyday decisions, operational resilience must become part of the broader conversation about trustworthy AI.
We cannot afford to focus exclusively on building intelligent systems without considering the infrastructure and organizational preparedness required to sustain them.
The future of trustworthy AI will depend not only on how intelligently our systems operate, but also on how responsibly we prepare for the moments when they cannot.
So, here is my question to technology leaders, educators, researchers, and policymakers:
If the AI-powered systems your organization depends on became unavailable tomorrow, would your existing governance strategy be enough to keep essential operations running?
Lola Longe, Ph.D.
Founder, SAFER AI Protocol LLC
