Artificial intelligence in Switzerland: what’s new for 2025

Among the changes in Switzerland in 2025 will be stricter rules for social platforms which make heavy use of AI algorithms. Keystone / Urs Flueeler

New regulations, a “Swiss ChatGPT”, and autonomous vehicles: 2025 marks a pivotal year for the development of artificial intelligence (AI) in Switzerland.

Last year, AI development accelerated considerably in the Alpine country. At the World Economic Forum (WEF), the Swiss government underscored its commitment to positioning the nation as a leader in responsible, inclusive and transparent AI. Driven by the EU’s AI Act, Swiss federal authorities have said they plan to reveal a regulatory proposal this year following delays. The legislation aims to minimise risks such as discrimination and surveillance.

Additionally, Switzerland plans to introduce a digital platform law to combat disinformation, deepfakes and violent content on social media platforms such as Facebook, X, Instagram and TikTok. It will reinforce users’ rights and demand more transparency on the content offered in the feeds of social media pages.

From a legal perspective, another major change this year is the green light for autonomous vehicles on Swiss roads. Could Switzerland be heading towards a driverless future?

On the technological front, the government and Swiss federal technology institutes are developing specialised language models for strategic sectors – a kind of targeted “Swiss ChatGPT”.

Michael Wade, professor at the business school IMD and director of the TONOMUS Global Center for Digital and AI Transformation, weighed in on these key changes to the Swiss AI landscape in 2025.

1. Towards Swiss AI regulation

The Swiss government has made it clear that AI regulation will be a central focus in the new year. In their Digital Strategy for 2025External link, authorities emphasised the importance of rules to protect fundamental rights and democracy from AI risks while promoting innovation and Switzerland’s competitiveness.

This isn’t the first time Switzerland has chosen to regulate specific technological areas: the country’s cryptocurrency lawExternal link is considered globally advanced. However, Swiss authorities hesitated for some time before finally deciding, at the end of 2023, to issue specific AI regulations – likely influenced by the imminent implementation of the EU’s AI Act.

A legislative proposal is expected to be published early this year, according toExternal link Communications Minister Albert Rösti, although it is delayed compared to initial announcements. For Michael Wade, it’s better late than never: “Not regulating AI would be like allowing pharmaceutical companies to invent new drugs and treatments and release them to the market without testing their safety.”

2. Stricter rules for digital platforms

A new law will impose stricter rules on major social platforms such as Google, Facebook, YouTube and X. The legislation is expected to be unveiled early this year, according to Angela Müller and Estelle Pannatier of AlgorithmWatch CH, who are closely monitoring government consultations.

“We need to turn the tables and establish rules for these platforms to facilitate constructive public debate that benefits society and democracy,” said the two activists.

The major social media platforms are responsible not only for spreading informational and entertaining content but also for fake news, manipulated images and videos (deepfakes), disinformation, and incitement to hatred and violence. With AI integration, social media platforms are increasingly skilled at capturing attention and shaping public opinion, often following opaque rules set by private companies.

The EU has already addressed this through the Digital Services Act and Digital Markets Act (DMA), enabling member states to demand safeguards from social media platforms against harmful content. For instance, when Meta (which owns Facebook and Instagram) announced plans to end its fact-checking program in the US, the EU quickly reactedExternal link, warning that any similar action in Europe would require a risk assessment submitted to the European Commission.

Wade argues that companies should establish ethical guidelines to ensure the safe and responsible use of AI. “Unfortunately, many Silicon Valley companies are heading in the opposite direction,” he adds.

3. Green light for autonomous vehicles

Another major development involves autonomous vehicles, which will be authorised on certain cantons’ designated road segments starting March 1, 2025. On motorways, autopilot features can be used, provided a driver remains present behind the wheel.

Autonomous driving relies heavily on AI integration. Algorithms act as the vehicles’ eyes, brain, arms and legs, enabling them to perceive, navigate and react in various situations. However, full autonomy will not be allowed – only vehicles monitored by an external operator will be permitted.

The benefits seem numerous: improved road safety (since 95% of accidents are caused by human error) and greater mobility for older people or those with disabilities. However, the technology is not yet mature enough for city streets and busy areas with pedestrians and unexpected events.

“There’s a need for more investment and for tech companies to commit, but we’re not there yet in Switzerland,” Wade says. Globally, the US and China are leading the implementation of autonomous vehicles, with robotaxi services already available. Both nations have invested billions in developing this technology.

4. Working on Swiss versions of ChatGPT

Switzerland also aims to develop large language models (LLMs) for specific applications in fields such as science, education, healthcare, robotics and climate studies. These efforts won’t replicate a general-purpose model like ChatGPT but will instead focus on areas where Switzerland excels and can provide added value, such as healthcare and pharmaceuticals.

“We don’t need another generative AI model but rather models that are more accurate, cost-effective and environmentally friendly,” Wade explains.

The Swiss federal technology institutes are collaborating to free Switzerland from dependence on opaque systems developed and managed by private companies behind closed doors. “In Switzerland, we have the infrastructure and some of the world’s best talent: we can lead technological development in key areas of societal importance and align it with Swiss values,” says Antoine Bosselut, head of the Natural Language Processing Systems Research Group at the Swiss Federal Institute of Technology Lausanne (EPFL) and member of the steering committee of the Swiss AI Initiative.

The project is ongoing, with innovative models already developed for fields such as medicineExternal link. Upcoming releases, including a Swiss LLM and tools for biomedicine and meteorology, are expected by summer. However, the Swiss AI Initiative is unlikely to cause an immediate shift towards more reliable AI tools. Competing with the scale, reach and resources invested by countries such as the US in building large-scale AI models remains challenging, according to Wade.

“Switzerland, like most countries, still relies heavily on foreign technologies, particularly from the US. This will likely continue this year,” he concludes.

Edited by Veronica DeVore/ts