Skip to content
Briefings are running a touch slower this week while we rebuild the foundations.See roadmap
AI: Jobs, Power & Money
4APR

EU Digital Omnibus second trilogue collapses

2 min read
20:44UTC

The EU Digital Omnibus second trilogue ended on 28 April without a deal; the third is scheduled for 13 May with the binding employer AI literacy clause still contested between Council and Parliament.

EconomicDeveloping
Key takeaway

The EU has failed twice to agree the lightest workplace AI clause; 13 May is the next test.

The EU Digital Omnibus second trilogue ended on 28 April 2026 without a deal 1. The third trilogue is scheduled for 13 May 2026. The binding AI literacy clause, which would require employers to ensure workers understand the AI systems they interact with, remains contested between the European Parliament and the Council, with the Parliament pushing for a binding obligation and the Council preferring non-binding encouragement language. The 28 April session was flagged as the second trilogue in the run-up and originally placed on the early-April schedule .

Council and Parliament differ on a single procedural word, but the operational gap that word opens is wide. The literacy clause does not regulate AI deployment; it regulates whether the people interacting with AI in the workplace must be told what they are interacting with and how. The Council's preferred non-binding language would let member states decide whether to enforce the obligation. The Parliament's binding version would force employers to provide minimum AI training across the bloc. Neither version creates the kind of dismissal-protection precedent the Hangzhou Intermediate People's Court issued in days; even the literacy clause, if it survives, takes effect at a multi-year horizon.

The procedural map across the three jurisdictions tells the briefing's clearest story. Beijing's court ruled in December 2025; Hangzhou's appellate bench affirmed in April 2026; both rulings are citable in any Chinese labour court tomorrow. Brussels' second trilogue collapsed without a deal on a non-binding training obligation. Washington's most viable vehicle, the Warner-Rounds commission, would not deliver an interim report until autumn. Three regulatory routes run at three speeds, with speed inversely correlated to the political pluralism of the system.

European workers face the same protection regime they faced a month ago. The General Data Protection Regulation and the AI Act provide some protection against algorithmic decision-making in specific contexts; neither addresses AI-driven dismissal as such. The 13 May trilogue tests whether the EU's slowest legislative cycle can produce even the lightest workplace protection inside a calendar year.

Deep Analysis

In plain English

The European Union passes laws through a negotiation process between two bodies: the European Parliament, which represents voters, and the Council, which represents national governments. When they disagree, they hold meetings called trilogues to find a compromise. The Digital Omnibus is a large EU legislative package covering digital rules. One clause in it would require employers to ensure their workers understand the AI systems being used in their workplaces, such as AI tools that evaluate their performance or help decide their schedules. The second attempt to agree on the package, on 28 April 2026, failed. The third attempt is scheduled for 13 May. The specific sticking point is that clause on worker AI literacy: national governments (the Council) generally want it weaker or removed; elected MEPs generally want it kept or strengthened. If the clause survives into law, any company operating in the EU would have to train its workers to understand the AI tools affecting their jobs. This would apply to UK companies with EU operations too.

Deep Analysis
Root Causes

The literacy clause's controversy reflects a structural division in EU employment law that predates the Omnibus. The Works Council Directive and the Information and Consultation Directive already require employers to inform and consult workers on major changes, including technology changes. The literacy clause extends this to a positive obligation to ensure worker comprehension, moving beyond notification to a measurable outcome obligation.

Council opposition comes primarily from Germany, France, and Poland, each of which has industry lobbies arguing that the clause increases compliance costs without specifying what counts as adequate literacy. Parliament support comes from MEPs in countries with stronger works council traditions (Austria, the Netherlands, Sweden) who view the clause as a natural extension of existing consultation rights to the AI context.

The Omnibus timetable pressure also matters: both sides want the broader digital package finalised before the summer recess. Both the Parliament and Council have confirmed the literacy clause remains the unresolved item blocking the broader Omnibus package. A deal before summer recess requires one side to concede.

What could happen next?
  • Consequence

    If the literacy clause survives into the 13 May deal, UK companies with EU operations will face a new compliance requirement to document and demonstrate worker AI literacy programmes, likely requiring third-party assessment vendors.

    Medium term · 0.68
  • Risk

    If the trilogue fails a third time, the Omnibus may be split, with the literacy clause deferred to a standalone instrument, which would delay any binding EU AI worker protection by 12-18 months beyond the current timeline.

    Short term · 0.6
  • Precedent

    An EU binding AI literacy obligation for workers would be the world's first binding employer obligation specific to AI, creating a template that Japan, South Korea, and potentially the UK could adopt under their own digital legislation.

    Long term · 0.55
First Reported In

Update #8 · Beijing court bans AI sackings as Big Tech burns cash

European Parliament· 2 May 2026
Read original
Different Perspectives
TSMC and Taiwan chip supply chain
TSMC and Taiwan chip supply chain
Nvidia's 17% headcount growth to 42,000 on $81.6 billion in quarterly revenue depends on TSMC's CoWoS advanced packaging capacity constraining H100 and B200 supply, sustaining margins above 70%. The AI build-out's sole headcount-growth story runs through a Taiwan supply chain that has no parallel in downstream software.
Displaced tech workers globally
Displaced tech workers globally
CrowdStrike's SEC disclosure puts AI attribution on a material regulatory record for the first time, but Oracle's Massachusetts WARN clock expired unfiled after up to 14 workers were logged as remote despite office proximity. The legal apparatus cannot enforce what it cannot see: hybrid reclassification, GCC transfers, and hires never made.
UK workforce and policymakers
UK workforce and policymakers
ONS recorded UK vacancies at 705,000, below the pre-pandemic baseline for the first time, as payrolled employment fell 210,000 year on year with real wage growth at 0.1%. The Bank of England's AI worst case assumed 500,000 additional unemployed from a baseline above 730,000; the UK is already below that floor, and ONS still publishes no AI-exposure breakdown.
India IT workforce and graduates
India IT workforce and graduates
NASSCOM's FY2026 data shows net sector growth of 140,000, but entry-level hiring fell 20-25% as the growth concentrated in in-house GCC offices requiring mid-career specialists. Indian graduates who previously entered through TCS, Infosys and Wipro fresher programmes find that channel closing at both ends: outsourcers cutting and GCCs not hiring at the junior level.
IG Metall and European trade unions
IG Metall and European trade unions
European labour bodies see the market reward pattern, cuts on record revenue, as investor preference for short-term margin extraction over validated AI productivity. They note the EU Digital Omnibus provisional deal has dropped binding employer AI-literacy obligations at the precise moment the ILO-NASK index has quantified that 3.3% of global workers are in the highest AI exposure category.
Federal Reserve Board
Federal Reserve Board
Governor Cook told Stanford's SIEPR on 27 May that speculative-grade software bond spreads have widened on AI-disruption concern, moving AI displacement from a labour observation into the Fed's financial-stability mandate. The Fed cannot resolve structural labour transformation through rate policy, so Cook routed the concern through the one channel the Fed does control.