News
Poland's Sejm Finishes AI Law Vote, Bill Heads to President
On July 3, Poland's Sejm reviewed the Senate's amendments to the artificial intelligence systems law and adopted 24 of the 25 proposed changes, concluding parliamentary work on the bill. The law, which implements the EU AI Act, now goes to the president for signature ahead of the EU's August deadline.
Poland's Sejm (the lower house of parliament) concluded work on the artificial intelligence systems law on July 3, 2026, voting on amendments previously submitted by the Senate. Lawmakers adopted 24 of the 25 proposed changes, rejecting one by a vote of 405 in favor, 13 against, and 17 abstaining. This was the final step in parliament, the bill now heads to the president's desk.
The bill was submitted to the Sejm on April 9, 2026, as Poland's implementation of the EU's artificial intelligence regulation. After a first reading and work in the digitization committee, lawmakers passed the law in June by an overwhelming majority, after which the document moved to the Senate, which proposed its own amendments. July's vote on those amendments closes out the parliamentary stage.
What the law establishes
The centerpiece of the new legislation is the creation of the Commission for the Development and Security of Artificial Intelligence, known by its Polish acronym KRiBSI. It will serve as Poland's national market surveillance authority for AI systems and models, as well as the sole point of contact with EU institutions. The commission's chair is appointed by the Sejm with the Senate's consent for a five-year term.
The law also introduces regulatory sandboxes, controlled testing environments for innovative AI systems, which micro, small, and medium-sized companies can use free of charge. Any citizen who believes their rights have been violated by an AI system's operation will be able to file a complaint directly with KRiBSI, while appeals against the commission's decisions will be handled by the Court of Competition and Consumer Protection (Sąd Ochrony Konkurencji i Konsumentów) in Warsaw.
Penalties and leniency mechanism
Using banned AI practices carries penalties of up to 35 million euros or 7 percent of a company's global turnover, while less serious violations carry fines of up to 15 million euros or 3 percent of turnover. The law does include a negotiated leniency mechanism, though: an entity that voluntarily reports a violation and cooperates in remedying its effects can have its penalty reduced by as much as 90 percent. Proceeds from the fines will flow into the state budget, and enforcement will follow standard administrative enforcement procedure.
In the global technological revolution, we are not satisfied with the role of a tenant who merely pays rent and uses foreign solutions - Krzysztof Gawkowski, Minister of Digital Affairs
Pressure from the EU deadline
A hard deadline looms over the entire process: by August 2, 2026, European Union member states must have national oversight frameworks for the AI Act in place. By wrapping up parliamentary work three weeks ahead of that date, Poland finished with considerable room to spare compared with some other EU countries, which are still arguing over the shape of their own implementing legislation.
For companies operating in Poland, this ends a period of uncertainty over who will oversee the AI systems built into their products, and under what rules. Businesses using generative models in customer service, credit scoring, or recruitment will now have to reckon with an actual oversight body and a real possibility of inspection, rather than just a declarative obligation to comply with the EU regulation.
The law now awaits the president's signature. The standard deadline for signing it, referring it to the Constitutional Tribunal, or sending it back to the Sejm for reconsideration is 21 days, which leaves little room for further delay given the EU's August deadline.
Sources: Rzeczpospolita (rp.pl), CyberDefence24 (cyberdefence24.pl), Prawo.pl (prawo.pl)
