My article on Infosperber calls on Switzerland to take an active role in shaping digital policy—especially in the sensitive areas of surveillance and control.

Switzerland, too, should set limits on the surveillance state

On Wednesday, June 14, 2023, theEuropean Parliamentadopted the framework for anAI Regulationon artificial intelligence (AI). The vote was clear: 499 in favor, 28 against, with 93 abstentions. The EU strategy takes a risk-based approach and aims to regulate and restrict the use of AI more strictly the greater the risk posed by a particular AI application. 

Where the use of artificial intelligence threatens to oppress people, such applications are subject to a total ban. These include:

• “Social scoring” systems that evaluate individual behavior. The surveillance policies in China, where individuals, government employees, companies, organizations, and associations can be classified and evaluated based on data, have a deterrent effect. Misconduct is disciplined and punished—for example, with reduced internet speeds or flight bans.
• Biometric recognition systems that monitor public spaces in real time—except when law enforcement agencies obtain a court order to investigate serious crimes.
• Biometric categorization systems that use sensitive characteristics such as gender, race, ethnicity, nationality, religion, and political orientation, and that may have discriminatory effects. 

Overall, the EU regulation aims to prevent violations of human rights and the right to privacy. Providers of AI models are expected to ensure robust protection of fundamental rights, health and safety, as well as the environment, democracy, and the rule of law. AI systems such as “ChatGPT” must also disclose that their content is AI-generated in order to meet transparency requirements.  

Unity sounds different

The European Parliament’s goals are ambitious. But the chorus of voices that must now find common ground in Europe is diverse and discordant. Prior to the European Parliament’s decision, the European People’s Party (EPP)—an alliance of conservative parties—had sought to place greater emphasis on terrorism. Bavaria’s Minister for Digital Affairs, Judith Gerlach, criticized the proposal, stating, “The current draft legislation stifles innovation and particularly harms our Bavarian economy, for which AI is a key technology.” She argued that the proposals to date urgently need to be revised. Smaller companies, in particular, would be unable to cope with the excessive regulation and bureaucracy. Similarly, the German Retail Association (HDE) expressed concerns about overregulation. While rules are necessary, they cautioned, one must not throw the baby out with the bathwater.

During the parliamentary debate, Axel Voss of the CDU called the proposed ban on real-time facial recognition in public spaces regrettable and a missed opportunity: He argued that AI, when used properly, could lead to significantly greater public safety. Svenja Hahn, the FDP’s digital policy spokesperson from the liberal camp, countered that facial recognition has no place in a liberal democracy when used for surveillance.

The disagreements—which, despite a compromise, were ultimately not resolved—demonstrate that June 14 marked, at most, a milestone in the debate. The final decision will be made by the member states’ representatives in the EU Council of Ministers. In a speech to business leaders, French Prime Minister Emmanuel Macron immediately announced that France would allocate seven billion euros in new funding for AI development. The aim is to prevent China and the U.S. from dominating the field. This is because there is great concern that AI development could bypass European countries due to strict regulations. Google, for example, is currently launching its chatbot “Bard” in most countries around the world. The EU was initially excluded due to data protection concerns. At the last moment, however, a rather vague compromise was reached.

The use of AI in migration policy remains controversial. NGOs point out that AI—ranging from autonomous drones to speech recognition software—is being used to secure borders against illegal crossings or to detect false statements in the asylum process. European Digital Rights (EDRi), a coalition of numerous European NGOs, criticizes the fact that the AI Regulation does not improve human rights, particularly those of migrants. The European Parliament has failed to introduce provisions that protect the rights of migrants in the face of increasingly discriminatory surveillance.

Switzerland as an observer on the sidelines

While the EU is forging ahead with regulations on a global scale, Switzerland—as is often the case with European policy issues—is proceeding at a leisurely pace and observing the situation from the sidelines. In apress releaseon April 18, 2023, the Federal Council stated that Switzerland is affected by European Union (EU) regulations in many areas. However, it does not see any immediate need for action. 

No significant barriers to market access are expected for Switzerland in the digital sector. This is also because the new mandatory regulations apply generally to all providers, not just those from third countries such as Switzerland.

The bottom line is this: Switzerland will not be able to avoid external influences. Swiss digital products will have to comply with EU regulations if they are sold in the EU. However, there is hope that existing laws can simply be amended. 

But laissez-faire is not a good solution in politics. Switzerland will have to comply with the requirements set forth in laws such as the EU Convention. Furthermore, Switzerland can hardly allow or promote developments that are considered extremely dangerous risks outside of Switzerland. 

Now that the outlines of a European digital policy are taking shape within the EU, Swiss policymakers should act quickly to formulate their own requirements to address these challenges—requirements that protect human rights and prevent widespread surveillance.

From: www.infosperber.ch