Ireland is having a moment in artificial intelligence that most American HR leaders have not clocked yet. A research note from the Expert Group on Future Skills Needs, the Irish government’s skills advisory body, published in late August, found AI-related jobs being created in Ireland at two to three times the pace of neighboring countries.
Close to 12% of Irish job postings now reference AI skills. In the U.S., that figure is 4%. In the U.K., it is 6%. Microsoft’s AI Diffusion Report puts Ireland second in Europe for overall AI adoption, behind only Norway, and large-enterprise AI usage here has climbed from 31.2% to 57.3% since 2021. A market of 5 million people is running ahead of the big ones.
We built TalentHunter inside that environment, and inside a regulatory one that got there before America did. Recruitment was classified as high-risk under the EU AI Act before we had our first paying client, and a candidate’s right to a plain explanation of a hiring decision was close to law before most vendors had thought about it. We joined the European AI Pact, the voluntary framework the European AI Office set up ahead of the Act’s deadlines, because it put us in the room where the standards were being worked out.
That environment shaped the product. Every résumé score a candidate receives can be sent to a human reviewer on request. Our terms of service state plainly that no candidate is ever automatically rejected by an algorithm. Our policies open in ordinary language before they open in clauses. Our client base spans hospitality, security, facilities, construction and healthcare staffing across Ireland, businesses that mostly do not have a dedicated compliance team, which meant the standard had to work at that scale or it did not work at all.
See also: The countries winning the global talent race
U.S.-based companies still have to abide by the EU AI Act if their AI system is used there
Here is the part that matters if you are sitting in Chicago or Dallas and hiring into Dublin or Berlin: the EU AI Act does not care where your company is headquartered. It applies to any organization whose AI system’s output is used inside the EU. Screen applicants for a role in your Cork office with a tool running on a server in Virginia, and you are a deployer under the Act, with the obligations that follow.
That means using the system according to the provider’s instructions, keeping a human genuinely in the oversight loop, telling candidates and workers that the system is in use and keeping the logs to prove all of it. Penalties scale to global revenue, the same design as the GDPR fines American firms already know.
The timeline has just shifted, and it is worth being precise, because the shift is being misread. The high-risk obligations for hiring tools were due to bite on Aug. 2 of this year. Under the EU’s digital omnibus package, that deadline has moved to December 2027. Some vendors are treating the delay as a reprieve. It is a short one. The Act’s transparency rules still took effect this August, the GDPR’s Article 22, which restricts fully automated decisions about people and gives candidates a right to human review, has been in force the whole time, and the recruitment tools employers buy this year will still be in service when the high-risk rules land.
What compliance actually required of us was not a document produced at the end. It was an audit trail that exists from the first decision, and that is not something anyone can retrofit.
America is heading toward the same place through a different route, city by city and state by state. New York City requires an independent bias audit and candidate notice before an automated tool is used on anyone residing there. Illinois amended its Human Rights Act from Jan. 1 of this year, so employers must disclose when AI is used in an employment decision, and using AI that discriminates is a civil rights violation. California’s civil rights regulations, in force since October 2025, treat automated decision systems like any other employment practice, and whether anyone ran bias testing counts as evidence in a discrimination claim. Colorado’s sweeping 2024 law was blocked by a federal court in April and replaced in May with a narrower notice-based statute taking effect in January 2027. Underneath the patchwork, the federal picture is instructive.
The EEOC withdrew its guidance on AI hiring in early 2025, but Title VII, the ADA and the ADEA did not move. They apply to an algorithm exactly as they apply to a person, and “the algorithm did it” is not a defense.
If you are evaluating AI hiring tools against that patchwork, four questions will tell you most of what you need to know. Can a person point to the exact moment they, not the algorithm, made the call? Can a rejected candidate get a real explanation in language they understand? Can the vendor hand you the documentation behind that answer, the bias testing, the logs, the audit trail that New York already demands and other states are drafting toward? And is fairness something the product does or something the marketing page claims? We have been answering those questions under a regulator’s eye since before we had revenue. American employers will be asking them of every vendor soon enough.
The post What building fair hiring under real regulation taught us, and why the U.S. is next appeared first on HR Executive.
This article was originally published on HR Executive. Click below to read the complete article.