The question comes up in almost every adoption project, usually before the first system goes live: is there an AI law in Switzerland that we have to comply with? The answer is short. There is no overarching Swiss AI law. On its topic page, the Federal Office of Justice states that Switzerland currently has no specific legislation on AI. The Federal Chancellery and OFCOM put it the same way: no overarching legislation specific to AI, with adjustments made where necessary in the individual sectors.
It does not follow that no law applies to AI projects. It only follows that there is no dedicated law for AI.
What is binding today
What is binding is the law in force. Data protection law, unfair competition law, copyright law and the relevant sector-specific law apply unchanged, even when a decision is prepared by a model rather than by an employee. There is no exception for systems declared as pilots.
In addition, there are supervisory requirements for regulated industries. On 18 December 2024 FINMA published Supervisory Notice 08/2024 “Governance and risk management when using artificial intelligence”. It is addressed to institutions supervised by FINMA. For all other companies it creates no obligations.
In the other direction: AI-specific obligations of the kind known from EU law do not exist in Switzerland today. There are no statutory risk classes, no conformity assessment, no registration requirement and no legal obligation to provide AI competence training. Any further expectations in circulation stem from self-regulation, industry dialogues or the law of other jurisdictions.
Which regulatory approach the Federal Council decided on
On 12 February 2025 the Federal Council set the approach. The Council of Europe’s AI Convention is to be adopted into Swiss law. The legislative adjustments needed for this should be made on a sector-specific basis wherever possible. General, cross-sector regulation is limited to central areas relevant to fundamental rights, such as data protection.
The Federal Department of Justice and Police (FDJP) was tasked with drafting a consultation proposal for new rules on the use of AI. The content named includes transparency, data protection, non-discrimination and supervision. In parallel, the Department of the Environment, Transport, Energy and Communications (DETEC) is to propose legally non-binding measures, such as industry solutions and voluntary commitments. Regulatory work in individual areas such as healthcare or transport continues alongside this.
What is to be expected, therefore, is not a single Swiss AI law regulating everything, but amendments to several existing enactments.
What has happened since the signing of the Council of Europe Convention
Federal Councillor Albert Rösti signed the Council of Europe Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law on behalf of Switzerland on 27 March 2025 in Strasbourg. According to the Council of Europe, the Convention is the world’s first legally binding international treaty in this field and was opened for signature on 5 September 2024.
Signed is not ratified. The signature creates no directly applicable obligations for companies. The Convention takes effect only through the national legislative project.
The work on this is being carried out by the Federal Office of Justice, together with OFCOM, the Directorate of International Law and other federal bodies. The Federal Office of Justice has set up an advisory group of internal and external experts and, on 27 October 2025, held a participatory working session with around 60 representatives of various interest groups. At its meeting of 28 October 2026, the Federal Council took note of the interim status, including OFCOM’s report “Sector-specific measures for the regulation of artificial intelligence” and a “Guide to an AI code” with draft text on ten topics in three levels of detail for associations and companies. OFCOM is to establish, by the end of March 2028, an internal administrative exchange of knowledge on the industries’ self-regulation activities and to further develop the guide. Individual offices are already conducting industry dialogues, for example the Federal Office of Culture with the “AI Dialogue Culture” since November 2025.
When the consultation is coming
The original mandate called for a consultation proposal by the end of 2026; as of retrieval on 27 October 2026, this still appears on the pages of the Federal Office of Justice, the Federal Chancellery and OFCOM. The reporting on the Federal Council meeting of 28 October 2026, by contrast, names spring 2027 for the opening of the consultation.
A second project is running separately. Following the adoption of Motion 24.4596 Gössi, the Swiss Federal Institute of Intellectual Property is drafting sector-specific rules for AI in copyright law. The draft first goes to the Federal Council, which decides on opening the consultation. According to a presentation by an IPI staff member on 26 October 2026, the project rests on three pillars: clear exclusive rights, extended collective licences and more transparency. Also mentioned are a remuneration claim against foreign AI providers and a reversal of the burden of proof in favour of rights holders; a public consultation could take place in spring 2027. Anyone training models on third-party content, or having them trained, should keep an eye on this proposal. It is not part of the Federal Office of Justice’s Convention project.
Does the EU AI Act apply to Swiss companies?
The EU AI Regulation entered into force on 1 August 2024. Whether it covers a Swiss company is determined by its own scope of application and not by Swiss law; the assessment is to be made against the text of the Regulation. With Regulation (EU) 2026/1744 of 8 July 2026, the European Parliament and the Council amended the AI Act with a view to simplifying implementation. What this changes and what remains unchanged, we set out in EU AI Act: Postponed is not cancelled. Many of the deadline overviews found online date from before the final text and are outdated.
May we continue our AI projects?
Yes. The absence of an AI law is neither a reason to wait nor a free pass. The useful question is not whether a project is permitted, but whether the company can later demonstrate what it did.
Regardless of the state of legislation, a company should be able to answer three points:
- which AI systems are in use and who is professionally responsible for them,
- which data and which decisions each system touches,
- how quality was checked before going live.
Whoever has these foundations can read a consultation proposal and assess within a manageable time what it means for their own organisation. How to set this up without slowing down ongoing projects, we describe in Introducing AI responsibly; we draw the line between obligation and expectation with our clients in Artificial Intelligence Advisory.
For the board of directors a timing question remains: anyone who will be affected by the proposal should follow the consultation as soon as it opens, and not only once it has been decided.