What is Switzerland’s position on AI regulation?

read aloud

Everyone seems to agree that there should be limits to what artificial intelligence (AI) can do. The European Union and the Council of Europe are currently developing a set of regulations. what about switzerland?

This content was published on July 9, 2023

July 9, 2023

Christian Lahloub

From social media, healthcare, and chatbots to semi-autonomous cars, artificial intelligence (AI) is a part of our daily lives, whether we like it or not. Now, the chapter’s top-level representatives recently sounded the alarm about the dangers of AI in an open letter. Risk mitigation must be a “global priority,” they wrote.

Angela Muller, head of AlgorithmWatch CH, explains that concerns revolve around human rights, the rule of law and democracy.External link, a civil society organization that critically monitors the development of AI. The potential negative impact of AI is already visible, she said, pointing to the Dutch childcare benefit scandal.External link Algorithmic racism is involved. In Mueller’s view, incidents of this kind have made AI regulation “relatively urgent.”

Federal Communications Agency Deputy Director Thomas Schneider agrees. “Data is the new oil and AI systems are the new motor,” he says. He added that people are aware that these are serious problems and that corresponding solutions need to be found.

Despite this, Schneider is also chairman of the Council of Europe’s Committee on Artificial Intelligence (CAI).External linkbelieves it is necessary to distinguish between workable and immediate solutions, which is why Switzerland is taking a somewhat wait-and-see attitude.

So far, no country in the Organization for Economic Co-operation and Development (OECD) has introduced AI-specific regulations.The most advanced institution in this regard is the European UnionExternal linkthe European Parliament agreed to the first draft of the proposed AI Bill (Artificial Intelligence Law) on 14 June.

These laws also cover high-risk applications, such as banning real-time facial recognition (as used, for example, in China’s social credit system).External link) or about language-assisting children’s toys that may encourage risky behavior.

As a ‘defender of human rights’, the Council of Europe likewise feels an obligation to develop its own legal mechanism on AI.External link. Müller stresses that the Council of Europe’s framework AI treaty does not conflict with EU regulation, but rather complements it. “Both sides follow different approaches and the EU wants to control AI through product safety.”

no overarching law

What is clear is that there is no single, comprehensive law covering AI. Schneider compares AI to engines, which are subject to specific laws and regulations depending on their use. The same is true for AI, which involves completely different applications.

“If a music-streaming service requires an AI system, the same algorithm will have a different impact than recommending the next step a surgeon should take during heart surgery,” Schneider said. said.

So a single comprehensive law is not enough, a fact the EU is aware of, he said. “There are about 30 proposals in the EU that deal exclusively with the digital sector,” Schneider explains. A key question for him is what aspects of the current law do not cover.

Mueller agrees. “We are not currently floating in a legal vacuum. We already have laws in place, including constitutions and protections of fundamental rights.” It’s about closing the gaps created by the challenges we face, she says.

But the situation is equally clear for her. “It’s not a one-size-fits-all law. It won’t work.” fields of law, i.e. a huge legal area.

Where is Switzerland located?

AI regulation is currently being debated in many countries, so what approach should non-EU Switzerland take? First of all, Switzerland, which has been a member of the Council of Europe since 1963, is considered to follow convention.

In addition, the United States, Canada, Japan, Israel, Mexico, etc. are also members of the Council of Europe, so their regulations could have far-reaching implications.

Thomas Schneider said Switzerland’s position at the moment is to wait and see and check different options. And it’s never alone. “Everyone is trying to see if what the EU is planning will actually work.”

In that sense, Switzerland is “one of the many typical countries that are not just sitting still, but analyzing and weighing their options without committing to anything concrete yet,” he said. . He added that once the process begins, it will take years, if not decades, of adjustments.

European patchwork?

Like the UK, Switzerland has not joined the EU since the UK left the EU. Will Europe finally face patchwork of her AI regulation?

AlgorithmWatch CH’s Angela Müller stresses that the aim of EU law is to try to prevent this, at least within the EU. “Nevertheless, the regulations put in place by the EU will always apply if an outside company wants to sell products within her EU,” she says.

Perhaps this also applies to companies in Switzerland and the UK. Moreover, there is no doubt that various industries exert some degree of political pressure to obtain legal securities in their respective fields.

Then there is the question of liability. Who’s to blame when an AI system breaks the rules? This is the topic of the upcoming US Transparency Conference.

Computers are never criminals, Muller said, because they have no criminal intent. Responsibility always falls on the person who develops the system or uses it for a specific purpose. Human actors should still be identifiable.

“If that becomes impossible, the core underpinnings of the rule of law will be undermined,” she concluded.

Edited by Barz Regendinger. Translated from German by Thomas Skelton-Robinson

Read more: SWI swissinfo.ch is accredited by the Journalism Trust Initiative