The EU AI Act classifies most recruiting AI as “high-risk,” which brings real obligations starting in 2027: documentation, human oversight, and conformity assessments before a system can be used in hiring across the EU. Here’s what to ask any screening vendor now, before the deadline is close.
Why recruiting AI is high-risk by default
The Act treats systems used for recruitment, candidate evaluation, and promotion decisions as high-risk because of their direct effect on someone’s access to employment. That classification applies whether the system fully automates a decision or meaningfully influences a human one.
What high-risk status requires
High-risk systems need a documented risk management process, human oversight built into the workflow rather than bolted on, and a conformity assessment before deployment. Vendors will need to provide documentation that a customer can hand to a regulator, not a marketing summary.
What to ask a vendor before 2027
Ask directly: do you have a conformity assessment in progress, what does your human oversight step actually look like in the product, and can you provide documentation a compliance team can review. A vendor without clear answers to these three questions is a bigger risk than the fine print of any single feature.





