BAG Rulings and EU AI Act Box In HR Automation as Tech Giants Slash Jobs Anyway

Even as Oracle cuts 21,000 staff and Cloudflare fires one in five workers despite record revenue, German courts and European regulators have effectively outlawed the one tool many employers once hoped would streamline such decisions: fully automated firings. A pair of recent Bundesarbeitsgericht (BAG) decisions, combined with the EU’s incoming AI Act, mean that any algorithm that decides a worker’s fate faces legal obstacles that few current systems can clear.

The first ruling, handed down on 1 April 2026, tightens the screws on mass-dismissal procedures. Any mistake in the mandatory notification to the authorities—such as filing it before concluding employee consultations—automatically invalidates every affected termination. The BAG explicitly aligned itself with the European Court of Justice, insisting that such procedural defects cannot be cured retroactively. For HR software handling hundreds of layoffs, that means juggling overlapping deadlines and consultation windows without a single slip.

A second judgment, dated 18 June 2026, strengthens parental leave protections. The shield against dismissal now applies separately to each parental leave period and kicks in as early as eight weeks before that period begins—even during the probationary phase. HR algorithms that fail to track individual parental status and the associated blackout dates in real time risk producing invalid notices.

The European Union’s AI Act adds another layer. Systems that determine employment conditions are classified as high-risk, with full compliance required by 2 December 2027; initial labelling obligations start as early as August 2026. Meanwhile the GDPR already bans fully automated decision-making unless the worker gives explicit consent. The European Court of Justice clarified that even calculating probability scores—such as a “risk of leaving” or “predictive performance rating”—can count as an automated decision. Penalties run up to €35 million or 7% of annual global turnover.

While EU regulators tighten rules on automated decisions, UK employers must also stay on top of their health and safety compliance obligations. Failing to have the right documentation can lead to significant fines. Avoid compliance gaps with a free Health & Safety at Work Act 1974 Toolkit. It includes nine essential tools—risk assessments, checklists, and a director's liability guide—to help you meet legal requirements and protect your workforce. Download the free Health & Safety at Work Act 1974 Toolkit

None of this has stopped the tech sector from shedding jobs. In May 2026 layoffs hit a new peak: Oracle let go 21,000 (13% of its workforce), Meta cut 8,000, Google dismissed 3,000, Cisco removed 4,000, and Cloudflare shed 20% of staff even while posting a record $639.8 million in quarterly revenue. The pain concentrates among new entrants: PwC found that US entry-level jobs have shrunk 35%, and globally, AI-exposed entry positions fell about 10% between 2019 and 2025. Germany’s Ifo Institute reports that 54.5% of domestic firms now use AI; in retail, nearly one in three academic jobs is deemed replaceable.

Yet the technology itself fails the compliance test. A study by Aithos concluded that no current AI model achieves an acceptable level of legal conformity. Experts warn against fully autonomous decision-making and instead recommend embedding AI as a support tool within existing human-led processes. Analog interfaces remain particularly tricky: for severely disabled workers, the integration office must consent to a dismissal, with a two-week window that cannot be compressed into a fully digital workflow. Without bridging that gap, the algorithm remains a liability, not a solution.

en | boerse | 69617872 |