Switzerland - AI Convention Ratification

Overview

On 12 February 2025 the Swiss Federal Council decided in principle that Switzerland will ratify the Council of Europe Framework Convention on Artificial Intelligence, Human Rights, Democracy and the Rule of Law (the AI Convention) and instructed federal departments to prepare the adjustments to Swiss law and accompanying non-legally binding measures necessary for implementation. The decision was based on a series of baseline analyses and sectoral reviews and aligns Switzerland with the Council of Europe’s effort to establish a legally binding, human-rights-centred framework for activities across the lifecycle of AI systems. Switzerland subsequently signed the Convention in Strasbourg on 27 March 2025. The Federal Council’s working approach emphasises sector-specific legislative change where feasible, horizontal regulation limited to core areas affecting fundamental rights (for example, data protection and non-discrimination), and complementary non-binding instruments to support compliance and innovation. Official Swiss publications summarising the decision and next steps are available from the Federal Department of the Environment, Transport, Energy and Communications (DETEC) and other federal offices; see DETEC press release (12 Feb 2025) and DETEC press release (26 Mar 2025). The Council of Europe’s overview and the Convention text are available via the Council of Europe digital resources and the European Commission/EUR-Lex documentation.

Definitions

The Convention and Switzerland’s working documents use standard lifecycle and rights-based terms: "artificial intelligence system" (AI system) — systems that, during their lifecycle, process inputs and produce outputs using computational models; "activities within the lifecycle" — design, development, testing, deployment, use, monitoring and decommissioning; "risk-based approach" — prioritising measures proportional to the severity and likelihood of adverse impacts to human rights, democracy and the rule of law; "party/implementing authority" — the national authority or authorities designated to oversee Convention compliance; and "high-risk uses" — uses that present significant threats to human rights or democratic functioning. Baselining documents prepared for the Federal Council provide Swiss-language legal mappings and analyses referenced in the decision; see Bakom / OFCOM AI overview.

Governance and Institutional Framework

The Federal Council mandated inter-departmental responsibilities: the Federal Department of Justice and Police (FDJP) leads drafting of enabling legislation, in collaboration with DETEC and the Federal Department of Foreign Affairs (FDFA). DETEC, FDJP, FDFA and the Federal Department of Economic Affairs, Education and Research (EAER) will prepare non-legally binding implementation measures. At the federal-administration level the Federal Chancellery (Digital Transformation & ICT steering) coordinates digital strategy and internal AI use policies. The Convention itself requires Parties to establish or designate one or more effective domestic oversight mechanisms for monitoring compliance; Switzerland’s practical interpretation will determine whether this takes the form of an existing regulator’s expanded remit (e.g., OFCOM, FOJ, or a new dedicated body) or a federated model with sector regulators (for example FINMA for finance or FOPH for health) assigned compliance tasks. Official Swiss pages describe responsibilities and the interdepartmental structure; see Federal Chancellery AI page and DETEC press release.

Key Focus Areas

Switzerland’s implementation planning is expected to concentrate on: (1) Protection of fundamental rights — ensuring AI deployment does not undermine human dignity, privacy or non-discrimination; (2) Transparency and access to information — measures enabling affected persons to understand decisions that affect them; (3) Accountability and oversight — risk documentation, reporting and domestic supervision; (4) Risk management and safety — lifecycle risk assessments, pre-deployment testing where appropriate, and post-market monitoring; (5) Data protection alignment — ensuring that horizontal data protection rules (the Federal Act on Data Protection and related ordinances) integrate with AI-specific obligations; (6) Sector-specific regulation — health, transport, finance and public procurement given their immediate exposure to impactful systems; and (7) Measures to foster safe innovation — sandboxes, certification, guidance and voluntary industry standards. The Convention formalises a lifecycle approach that addresses design and testing, deployment, continuous monitoring, and decommissioning, and invites Parties to prohibit or impose moratoria on uses judged incompatible with human rights or democratic processes. These focus areas align with the Federal Council’s stated objectives of preserving innovation while strengthening protections. See explanatory material and Convention text at the Council of Europe and EUR-Lex references: Council of Europe AI resources and EUR-Lex Convention documents.

Implementation Framework

Procedurally, the Federal Council’s decision requires (a) the FDJP, in collaboration with DETEC and FDFA, to prepare a consultation draft implementing the Convention by the end of 2026; (b) DETEC, FDJP, FDFA and EAER to prepare a non-binding implementation plan (industry measures, declarations, guidance) by the same deadline; (c) engagement with sectoral regulators to draft targeted amendments where sector law interacts with AI applications; and (d) public consultation and parliamentary processes that will determine the final transposition approach. The implementation framework will likely identify critical legal nodes for horizontal intervention (e.g., data protection, anti-discrimination, administrative law and procedural safeguards for automated decision-making) while leaving other domains to sectoral law. Until transposition, existing law applies. See the Federal Administration summary page for ongoing responsibilities and next steps at Bakom / Federal Administration AI overview.

Monitoring and Evaluation

The Convention establishes an international monitoring and cooperation mechanism; Parties are expected to report domestically and to cooperate multilaterally. Domestically, Switzerland will be required to set up oversight mechanisms to track compliance, collect risk and impact documentation, and provide periodic reporting. Monitoring will include sector-by-sector assessments, outcomes of non-binding measures, and periodic reviews of legal gaps. Evaluation metrics will likely combine qualitative human-rights assessments, incident reporting, risk register audits, and sector-specific safety metrics. The Federal Council’s mandate to produce an implementation plan for non-legally binding measures suggests an evaluative loop where soft-law instruments are assessed for effectiveness and adapted. Relevant Swiss guidance pages and baseline analyses referenced by the Federal Council provide methodology for monitoring: DETEC (12 Feb 2025).

Penalties, Liability, and Appeals

The Council of Europe Convention requires access to remedies for affected individuals and contemplates that Parties provide appropriate civil, administrative or criminal measures to address breaches, but it leaves detailed sanctions and liability rules to national implementation. Switzerland’s Federal Council decision does not fix penalties immediately; instead it requires implementing legislation which will determine the shape and scale of administrative fines, civil liability regimes, and potential criminal sanctions for grave breaches. Swiss sectoral regulators (such as FINMA for financial markets) are expected to integrate compliance and enforcement measures within their existing sanctioning frameworks. The Federal Council has signalled that the approach will preserve the rule of law and procedural safeguards, keeping in view possible direct democratic controls (parliamentary approval and potential referenda). See Convention remedy principles in the Council of Europe documentation and Swiss implementation planning at Council of Europe and DETEC.

Relationship to Other Instruments

The Convention is designed to be interoperable with other regulatory frameworks, notably the EU AI Act (Regulation (EU) 2024/1689) and existing human-rights and data protection instruments. For Switzerland, the Convention will interact with the Federal Constitution, the Federal Act on Data Protection (FADP), sectoral statutes (healthcare, transport, financial market law) and administrative law principles. The Federal Council’s decision explicitly referenced the need to ensure compatibility with trading partners’ approaches and to consider the EU framework when defining the scope of horizontal measures. The European Commission and Council documents provide comparative analyses and highlight how the Convention complements the AI Act; see the EU/EUR-Lex explanatory materials and the Council of Europe texts: EUR-Lex and Council of Europe.

International Alignment

The Council of Europe Convention is open to non‑member states and many non-members (including the EU, United States, Canada, Japan and others) have already signed or expressed interest. Switzerland’s ratification aligns it with an emerging pan‑Atlantic approach prioritising human rights and democratic protections in AI governance while leaving room for national implementation choices. The Convention’s monitoring and cooperation mechanisms are intended to support interoperability and mutual learning across jurisdictions. Switzerland’s approach (sector‑specific transposition, limited horizontal rules on core rights, complementary non-binding instruments) aims to reconcile regulatory alignment with preservation of national innovation capacity. See the list of signatories and explanatory notes in EUR-Lex documentation and Council of Europe resources: EUR-Lex and Council of Europe.

Implementation Timeline

Event	Date
Council of Europe Committee of Ministers adopts Convention	2024-05-17
Convention opened for signature (Vilnius ministers’ conference)	2024-09-05
Federal Council decision in principle to ratify	2025-02-12
Switzerland signs the Convention (Strasbourg)	2025-03-27
FDJP/DETEC/FDFA to prepare consultation draft implementing Convention	By 2026-12-31
DETEC/FDJP/FDFA/EAER to draft non-binding implementation plan	By 2026-12-31
Parliamentary procedure & potential referendum	Dependent on legislative calendar after consultation

Compliance Checklist

Item	Notes
Identify whether AI applications fall within lifecycle activities	Map systems and uses across design, deployment, operation and decommissioning
Conduct risk assessments and document mitigation	Maintain risk registers and evidence of mitigation
Ensure transparency & user information	Provide meaningful explanations for automated decisions
Review data protection alignment	Confirm FADP compatibility and update privacy notices
Adopt non-discrimination safeguards	Run bias testing and implement remedial controls
Prepare testing & validation records	Pre-deployment testing and post-deployment monitoring logs
Designate internal compliance lead	Nominate responsible person(s) for documentation and liaison with authorities
Engage with sector regulator guidance	Monitor FINMA, FOPH, FOCA, OFCOM guidance and sector-specific measures

Sources and References

Source	Type
AI regulation: Federal Council to ratify Council of Europe Convention (DETEC press release, 12 Feb 2025)	Primary Source
Switzerland signs Council of Europe Convention on Artificial Intelligence (DETEC press release, 26 Mar 2025)	Primary Source
Council of Europe – Artificial Intelligence / Convention resources (text, explanatory report)	Primary Source
EUR-Lex – Documents and annexes containing the Convention text and explanatory materials	Primary Source