What Montesquieu Can Teach Us About Governing AI
Over the years, France has given the United States much for which we should be grateful — from military support that helped secure our independence, to the Statue of Liberty that has welcomed generations to New York Harbor, and Alexis de Tocqueville’s insights into our young democracy. But I would argue that France’s greatest gift to America was an idea: That power needs checks and balances.
Montesquieu was a French philosopher concerned with how to protect liberty and prevent the abuse of power. In his 1748 book, "The Spirit of the Laws,” he argued that power must be checked by power. America’s founders took note. James Madison drew upon Montesquieu’s ideas to shape the Constitution’s separation of powers and explain why no single branch of government should hold unchecked authority.
Today, this idea deserves renewed attention as we consider the power of artificial intelligence and the companies that create it.
The more powerful a technology becomes, the more important it is to build systems that hold it — and the companies creating it — accountable. AI’s extraordinary potential should strengthen our commitment to checks and balances, not diminish it.
For more than 25 years, my role at Microsoft has included overseeing compliance and the infrastructure that supports it. What I’ve learned over the years is that good intentions are not enough. People in any part of an organization can make mistakes. Organizations need internal controls, clear responsibilities, and people empowered to ask difficult questions.
With AI, these safeguards must extend beyond the walls of any one company.
As recognized last week by our industry, companies that develop and deploy AI need internal controls and external audits. That’s why Microsoft has long relied in part on our internal audit team to assess our AI safeguards. We also need independent evaluators who can test systems and challenge assumptions. But checks and balances should not stop there. We need competitive markets and a diverse and healthy ecosystem that give customers meaningful choices. And we need rules and laws that establish clear expectations and accountability.
None of these mechanisms are sufficient alone. Together, they can create a system stronger than any individual safeguard.
But the ultimate check and balance must include people who think and speak for humanity.
This is why I believe so fervently in listening to institutions whose missions reach beyond technology itself. Faith-based organizations that bring moral perspectives. Labor organizations that speak for workers. Human rights groups that stand up for human dignity. And academics and universities committed to sustaining human civilization. These voices must help shape decisions, not simply react to them.
Checks and balances aren’t the enemy of innovation. I’d argue that they help create the conditions in which innovation can endure. A world without rules or regulation may offer companies greater freedom in the short term. But a lack of guardrails will make public trust harder to sustain over time.
We shouldn’t have to choose between progress and protection. We need to build institutions that deliver both. This means defining the future we want to create and working together to help build it.