Switzerland's AI governance landscape is shaped not only by the EU AI Act and the Council of Europe's AI Framework Convention, but increasingly by the technical standards and frameworks emerging from the United States. Understanding how these interact is becoming essential for Swiss compliance professionals.
Switzerland and the United States have maintained strong economic ties since the 19th century. As of 2024, the U.S. is Switzerland’s second-largest trading partner after the European Union, with particular depth in finance, pharmaceuticals, and technology. That relationship now extends into the infrastructure of AI itself: a large share of Swiss businesses rely heavily on American technology and digital infrastructure, with many AI systems running on platforms provided by U.S.-based companies — Microsoft Azure, Amazon Web Services, and Google Cloud among them.
A different approach to AI governance
In July 2025, the United States signalled that it would pursue an innovation-driven policy on AI — with an emphasis on enabling innovation. This direction was reinforced with the publication of the White House AI Framework on 21 March 2026.
The U.S. approach does not rely on a single, legally binding regulatory framework. Instead, governance is shaped by a combination of voluntary standards, institutional guidance, and sector-specific requirements. The NIST AI Risk Management Framework (AI RMF 1.0) is the most prominent example: a structured methodology for identifying, assessing, and managing AI-related risks that carries no legal obligation but has been widely adopted across industries and borders.
Non-binding does not mean irrelevant
The non-binding nature of U.S. frameworks does not diminish their practical significance. Organisations operating internationally — including Swiss banks, insurers, and technology companies — often adopt U.S.-developed methodologies because they are embedded in the platforms and tools they already use. Model evaluation approaches, risk documentation templates, and monitoring practices developed in a U.S. context can become de facto standards simply by virtue of their prevalence.
As a result, technical norms that originate in Washington or Gaithersburg can shape how AI systems are built, assessed, and governed worldwide — including in Switzerland — without any formal legal mechanism.
Switzerland’s layered governance environment
Swiss organisations already face binding obligations from two directions. The EU AI Act imposes requirements on AI systems placed on or used in the EU market — directly relevant for Swiss companies with EU exposure. The Council of Europe’s AI Framework Convention, which Switzerland has signed, creates additional legally binding obligations around human rights, democracy, and the rule of law in the deployment of AI.
Alongside these, the technical practices used to develop and evaluate AI systems are often aligned with U.S. standards — creating a situation in which Swiss compliance teams must navigate obligations and expectations from multiple, distinct regulatory traditions simultaneously.
What this means in practice
The convergence of different governance approaches — from the non-binding U.S. framework to the more prescriptive EU regulatory structure — means that Swiss organisations cannot look to a single reference point for AI compliance. The legal requirements come primarily from Europe; the technical benchmarks often come from the United States; and the underlying infrastructure sits largely on U.S. platforms governed by their own terms and policies.
For compliance officers and risk managers in Swiss enterprises, this layered landscape demands a clear-eyed assessment of which frameworks apply, where they overlap, and where they diverge. It also reinforces the case for governance practices — such as systematic model documentation, ongoing performance monitoring, and structured risk assessment — that are robust enough to satisfy multiple frameworks at once.