EU AI Act Article 50 Labeling Deadline Is August 2: What US and Canadian Firms Delivering Work to European Clients Must Know

EU AI Act Article 50 Labeling Deadline Is August 2: What US and Canadian Firms Delivering Work to European Clients Must Know

A US accounting firm emails a transfer pricing memo to a Paris-based CFO. The memo was drafted substantially with AI. Starting August 2, 2026, six weeks from now, that memo may need a label.

That is what EU AI Act Article 50 means in practice for professional services firms with European clients — and most US and Canadian firms have not thought about it yet.

Two EU AI Act Deadlines. One Is August 2. One Is December 2027.

Before going further: there are two different EU AI Act compliance deadlines that matter to professional services firms, and they are commonly confused.

August 2, 2026 — Article 50 Transparency/Labeling: Firms using AI to generate content delivered to EU clients may be required to label that content as AI-generated. This is the deadline addressed in this article.

December 2, 2027 — Annex III High-Risk Systems: Firms using AI in high-stakes decision contexts (employment screening, creditworthiness assessment, legal interpretation at high stakes) must meet documentation and governance requirements. This deadline was extended from August 2026 to December 2027 via the EU Digital Omnibus package in May 2026. We covered that extension separately.

If you have already read about the "August 2026 EU AI Act deadline" in earlier coverage — including ours — that framing reflected what was known before the Digital Omnibus extension in May. The high-risk systems deadline moved. The Article 50 transparency deadline did not. August 2 still arrives.

What Article 50 Actually Requires

Article 50 of the EU AI Act imposes transparency obligations on "deployers" — companies that use AI systems to generate content they then deliver to others.

The EU Commission's Code of Practice on AI Labelling and Transparency, published earlier this year, defines the specifics:

• A uniform "AI" visual cue that must appear on AI-generated content delivered in EU jurisdictions
• Short explanatory text such as "Generated with AI" or "Manipulated with AI"
• Machine-readable metadata embedded in the file

The Code of Practice is technically voluntary for non-EU firms. That qualifier sounds reassuring until you read the legal analysis: Jones Day, Herbert Smith Freehills, and Bird & Bird — among the firms whose clients are asking them this question — describe the Code as "likely to become the key reference point for regulators and courts when assessing compliance" for non-EU entities whose output enters the EU market.

In plain terms: the label is not technically mandatory for a US firm, but the Code defines what a compliant label looks like, and EU regulators will use the Code when evaluating whether a non-EU firm's disclosure practices were adequate.

Who This Affects

This affects US and Canadian professional services firms — law firms, accounting firms, consulting firms, staffing firms — that deliver AI-assisted work product to clients in EU member states.

The scope question is whether your deliverables are "placed on the EU market" or "put into service" in the EU. The practical answer for most firms: if the deliverable goes to a client whose offices or primary operations are in an EU member state, it is in scope.

Common triggers:

• A contract review memo sent to a German-based company
• A transfer pricing analysis delivered to a French CFO
• A compliance gap report sent to a Dutch client
• An employment law analysis for a Belgian HR team
• A market entry report sent to an Italian company

Work product involving AI substantially generating the written content triggers the requirement. Work where AI edited or formatted a document a licensed professional wrote is a closer call — document your process.

UK clients are not in scope. The UK is not subject to EU AI Act requirements after Brexit.

The Practical Compliance Step

The simplest compliant approach is a disclosure line added to AI-assisted deliverables sent to EU clients.

The format does not need to be complicated:

This document was prepared with AI assistance and reviewed by [Attorney/CPA/Consultant Name], [Title]. Any AI-generated content has been reviewed and approved by the undersigned professional before delivery.

For law firms: this disclosure can sit alongside the existing engagement limitations language, or be added to the memo footer. For accounting firms: add it to the cover memo on AI-assisted analyses or reports. For consulting firms: add it to the executive summary page or document footer.

If you already have an AI disclosure policy for client-facing work — built for engagement letter compliance or internal governance — this extends that policy to EU-delivered work product with a format requirement.

Three Steps Before August 2

Step 1: Identify EU client work. Pull a list of active clients with EU-based operations or offices. These are your in-scope matters for Article 50.

Step 2: Audit how AI is used on those matters. For each EU client matter, identify whether AI is substantially generating written content in deliverables — memos, reports, contracts, analyses. If yes, those deliverables need the label.

Step 3: Create a disclosure template. Draft a standard footer or cover page disclosure statement. Route it through your bar association or AICPA guidance if available for your jurisdiction. Add it to your document templates for EU-facing work.

That is the compliance posture. It does not require a new tool, new hire, or new software. It requires a template and a process for applying it to EU-facing work.

What This Is Not

Article 50 is not the high-risk AI systems compliance requirement. You do not need to conduct a conformity assessment, establish a technical data log, or file documentation with an EU authority for Article 50. Those requirements belong to the Annex III high-risk path — and that deadline was extended to December 2027.

Article 50 is a content labeling requirement. It asks: did you tell your EU client that AI generated this content? That is the question. A clear, professional disclosure statement answers it.

The Bigger Pattern

Every bar association, regulatory body, and professional standards organization moving on AI right now is pointing in the same direction: when AI generates content that goes to clients, professionals are responsible for it, and clients have a right to know it came from AI.

Florida's Supreme Court made that point with its citation verification rule for court filings (effective June 15). California's State Bar has proposed six rule amendments requiring independent verification of AI output. The EU AI Act Article 50 makes the same point for EU-facing work product: disclose the AI.

The firms building disclosure practices now — as policy, not as one-off decisions — are building the governance layer that will be required everywhere within 18 to 24 months. Building it now for EU clients is the work. And it is six weeks away.

---

The EU AI Act's Article 50 transparency obligations take effect August 2, 2026. For the high-risk systems compliance deadline — extended to December 2, 2027 — see our coverage of the EU Digital Omnibus extension. For engagement letter disclosure language for US clients, see our AI disclosure engagement letter guide.