If your company uses AI to screen, rank, or filter job applicants in Europe, congratulations: you now need an annual third-party audit proving your algorithm is not discriminating against people. The EU AI Act's high-risk classification for AI hiring tools officially kicks in today, and the compliance requirements are not optional.

The rules: annual third-party bias audits, full technical documentation of how the AI makes decisions, human oversight mechanisms at every stage, and mandatory transparency disclosures to candidates explaining that AI was used in their evaluation. The penalty for getting this wrong: €15 million or 3% of global annual turnover, whichever is higher.

For a company like Amazon (which infamously scrapped an AI hiring tool in 2018 after it was found to penalize women), 3% of global turnover would be roughly $19 billion. That is not a slap on the wrist. That is an existential fine.

Why this matters beyond Europe

The Brussels Effect is real. When the EU mandated GDPR, every major tech company changed its privacy practices globally because maintaining two separate systems was more expensive than just following the strictest rules. The same dynamic will play out here.

Companies like HireVue, Pymetrics (now owned by Harver), and dozens of other AI hiring platforms will need to prove their models are not systematically disadvantaging candidates based on race, gender, age, disability, or other protected characteristics. And they will need to prove it to independent third-party auditors, not to themselves.

The timing is pointed. AI hiring tools have exploded in adoption. Companies use them to screen resumes, conduct video interviews, assess personality traits, and rank candidates. Some of these tools have been shown to penalize non-native English accents, downrank candidates with gaps in their resumes (which disproportionately affects women who took maternity leave), and favour candidates whose speech patterns match those of existing executives.

America is watching. America is not acting.

The contrast with the US could not be sharper. Congress has introduced 47 AI bills and passed exactly zero. New York City tried to enforce a local AI hiring audit law in 2023, and the result was widely considered toothless. Illinois has SB 3444 pending but it is stuck in lobbying warfare.

Meanwhile, the EU just did what it always does: wrote the rulebook while America argued about whether a rulebook was necessary.

The European Parliament is also debating broader AI data governance rules this month, with stakeholders pushing for more operational clarity on logging, documentation, and auditability of AI pipelines. Translation: this is not the last regulation. It is the first.

If you are a job candidate in Europe, your AI interviewer now has to explain itself. If you are a company using AI to reject people, you now have to prove you are not doing it because of who they are. And if you cannot prove it, €15 million says you should have tried harder.