Croatia in preparation for AI Law: Activists warn of risks to rights and call for safeguards going beyond EU AI Act

Politiscope event: AI and its impact on people and their rights

Politiscope recently hosted an event at the Croatian Journalists’ Association to discuss the human rights impacts of artificial intelligence (AI). The event, aimed at journalists and activists, comes at a crucial time as Croatia prepares a national law to implement the EU AI Act. With key decisions – such as the designation of the enforcement authority – still pending, this is a key moment to raise awareness of AI-related harms and influence future policies to incorporate human rights safeguards into national legislation.

Following the adoption of the EU AI Act in May 2024, the Croatian government has only recently formed a working group to draft a national law that will implement the Act.

Risks of AI: surveillance, discrimination, and unfair systems

The event opened with a powerful reminder that technology can harm people’s everyday lives if not properly regulated. Vanja Skoric (Program Director, EDRi member ECNL) highlighted how AI can threaten privacy, equality, and other rights. Ella Jakubowska (Head of Policy, EDRi) introduced Protect Not Surveil, a coalition working to ensure safeguards for people on the move against harms from technology such as AI, and Reclaim Your Face, a Europe-wide campaign demanding a ban on biometric mass surveillance. Participants were shown how biometric surveillance systems are often inaccurate, intrusive, and disproportionately affect marginalised communities.

Other speakers shared real-world examples of harm caused by AI systems. Jelle Klaas (PILP) described how strategic litigation in the Netherlands led to the removal of discriminatory algorithms used by banks for terrorism profiling. Filip Milošević (SHARE Foundation, an EDRi member) shared how civil society successfully stopped the rollout of biometric surveillance in Serbia. Nađa Marković (A11 Initiative) presented troubling findings on how automated systems unfairly deny social benefits to vulnerable people. Marija Renić (Ekran i Osmijeh) explained how social media algorithms subtly shape parenting practices. Tamara Zavišić (ETIK.AI) warned that corporate definitions of “values” and “safety” built into AI models often lack transparency and accountability.

Instead of the usual narratives of techno-solutionism and techno-geddon, the Politiscope event focused on shedding light on actual risks and harms to fundamental rights happening today. By raising awareness of algorithmic bias, discrimination, and unlawful surveillance, the event marked the first step in highlighting the urgent need for a collective push for stronger protections and accountability.

Ways forward: Embedding human rights perspective into AI governance

At the event, Duje Prkut (Politiscope) introduced the EU AI Act and explained how Croatia’s upcoming national law will shape the future of AI governance. A panel featuring experts from the Croatian Data Protection Agency (Anamarija Mladinić), SHARE Foundation (Danilo Krivokapić), Danes je nov dan (Maja Cimerman), and Politiscope (Duje Kozomara) discussed the importance of going beyond the EU AI Act’s minimum requirements.

Maja Cimerman presented DJND’s Register of Public Use of AI. This tool sets a higher bar for transparency by documenting where and how AI is used in public institutions, enabling public oversight and accountability. Initiatives like this show that proactive, citizen-centered solutions are possible and effective.

Politiscope offered the following recommendations for Croatia’s AI governance:

• public register of all AI and algorithms used by public and state administration
• ban on live biometric surveillance in public spaces
• meaningful participation of civil society in the development of the national AI legal and policy framework
• designating the DPA as the single market surveillance authority to enforce the AI Act in Croatia

Democratising AI Policy: Demands for openness and participation

Politiscope has submitted its recommendations to the working group under the Ministry of Justice, Judiciary, and Digital Transition, which is responsible for drafting the implementing act for the EU AI Act. Neither Politiscope nor any other civil society organisation (CSO) has been included in this working group. Representatives from the Data Protection Authority, the Ombudsman, and the leading private sector AI association, CroAI, have all expressed support for including Politiscope and other CSOs in the working group.

Legislative and policy-making processes in Croatia are becoming increasingly centralised and opaque. While private sector associations and trade unions are often included in working groups due to adherence to social dialogue principles, CSOs remain the most under-represented sector. Activities are underway to promote and ensure CSO participation in decision-making processes related to digital rights, including the AI Act. Politiscope will continue pushing for more inclusive and transparent legislative and policy processes.

The thematic conference organized by Politiscope was the first AI-specific live event in Croatia organised by an NGO or a watchdog organisation. It was a modest first step in bringing together civic stakeholders and journalists interested in safe, transparent, and ethical development and use of AI.

Contribution by: EDRi affilate, Politiscope