Poland's bold AI watchdog breaks from EU norms with new oversight body
Poland is setting up a new body to oversee artificial intelligence, breaking from the approach taken by most EU countries. The Commission for the Development and Security of Artificial Intelligence, or KRiBSI, will act as the sole market watchdog for AI systems in Poland. Unlike many member states, which rely on existing regulators, Poland's government has chosen to create an entirely new institution.
The move comes as the EU AI Act, adopted in May 2024, requires all national authorities to operate independently and without bias. However, KRiBSI's placement within the Ministry of Digital Affairs has raised concerns about its practical autonomy. The original plan for KRiBSI was to establish it as a fully independent agency. Instead, the final structure embeds it within the Ministry of Digital Affairs. Officials justify this decision by pointing to a shortage of AI specialists in Poland, arguing that centralisation avoids competition for experts.
KRiBSI will function as a collegial body, with representatives from major existing regulators. It will oversee all AI systems in Poland, except for general-purpose models. The draft law includes legal safeguards to protect its independence, but critics question whether its position inside a government ministry will limit its freedom.
Poland's proposal also introduces binding individual opinions and an AI Social Council—features not required under EU law. These additions set the country apart from others like Lithuania, which has simply folded AI oversight into its existing communications regulator.
Across the EU, approaches vary. By early 2026, only nine of the 27 member states will have officially designated their national AI authorities. Seventeen countries have chosen to distribute oversight across multiple sectoral regulators rather than creating new bodies. Italy, for example, passed its implementation law in September 2025, well ahead of the EU's full enforcement deadline of August 2, 2026.
Meanwhile, the European Commission has proposed amendments to the AI Act, known as the 'AI Digital Omnibus'. If adopted, these changes could push back enforcement timelines for high-risk AI systems, further complicating national preparations. KRiBSI is set to become Poland's central authority for AI oversight, despite concerns about its placement within the Ministry of Digital Affairs. The country's approach contrasts with most EU members, who have either assigned AI duties to existing regulators or spread them across multiple agencies.
The EU AI Act's full implementation remains scheduled for August 2026, though proposed amendments may alter this timeline. By then, Poland's new system will be tested against the law's requirements for independence and impartiality.
