EU AI Act

The Act groups AI uses into tiers. A spam filter is low risk and largely untouched. A system that screens job applicants or scores someone for a loan is high risk, meaning documentation, human oversight and accountability duties. A few uses, such as social scoring by governments, are banned outright.

The Act applies across the EU, but each member state has to name a national authority to enforce it. Poland did that with a law signed in July 2026, which is why its minister could claim Poland was first in the EU to put oversight of the AI market in place.

For US readers it matters because American companies selling AI products into Europe have to comply, and several US state bills borrow its risk-tier structure.