Share with your CHRO
The EU AI Act’s August 2025 deadline for high-risk AI systems is forcing a compliance reckoning that most HR leaders haven’t fully priced in. Seyfarth Shaw attorneys Yana Komsitsky, Paul Whinder, and Georgia Hill Smith lay out which workplace AI tools qualify as high-risk under the Act, specifically those used in recruiting, performance management, and workforce monitoring, and what governance structures organizations need to build now. The piece also maps the overlap with GDPR and emerging US state-level automated decision laws, making the compliance surface considerably larger than most HR teams have budgeted for.
What this means for your business
Whether your organization has EU employees or not, this piece is relevant if your AI vendors do, because the Act binds developers and deployers across the supply chain. HR leaders who bought AI recruiting or performance tools without a formal vendor assessment process are now sitting on undocumented risk. The trait that decides exposure here isn’t company size or geography, it’s whether the AI system makes or materially influences decisions about individual workers, which is exactly what most modern HR platforms do.
The governance gap Seyfarth identifies is real and underappreciated. “High-risk” under the EU AI Act isn’t a vague category. It covers AI systems used in employment decisions with binding consequences for workers, including hiring filters, productivity monitoring, and promotion scoring. Organizations deploying these tools are required to maintain technical documentation, implement human oversight mechanisms, and conduct conformity assessments before the system goes live. Most enterprise HR teams inherited these tools through procurement cycles that never contemplated that level of scrutiny, which means the documentation simply doesn’t exist.
The piece’s framing, written by attorneys whose practice depends on companies facing exactly this compliance complexity, tilts naturally toward building internal governance infrastructure rather than questioning whether some of these AI tools deliver enough value to justify the overhead. That’s worth noting, but it doesn’t make the underlying legal analysis wrong. The deployer liability exposure the article describes is accurate, and regulators in both the EU and several US states are moving toward enforcement postures that will make “we didn’t know” an expensive defense. CHROs who treat this as a legal team problem rather than an HR operations problem will discover, too late, that the decisions already made are theirs to own.
The leading indicator to watch is vendor behavior. If your AI HR tool provider hasn’t sent updated data processing agreements or conformity documentation by mid-2025, that’s not a paperwork lag, it’s a signal about their compliance readiness. A vendor who can’t produce a risk classification assessment for their own product leaves you, as the deployer, holding the liability. That’s the renewal conversation worth having now, not after the deadline passes.
Based on reporting from What Every Employer Should Know About AI Governance Before The EU AI Act Deadline – Employee Rights/ Labour Relations, originally published 2026-08-04 01:34:00.
