Regulation Guide

EU AI Act for Marketers: What You Actually Need to Know

Most EU AI Act coverage is written for lawyers. This guide is written for marketing teams — what your tools must disclose, which practices are now banned, and what you need to do before August 2026.

Last updated: May 2026

TL;DR

What is the EU AI Act?

The EU AI Act is the world's first comprehensive law governing artificial intelligence. Passed in 2024, it classifies AI systems by risk level and sets obligations for the companies that build and use them. It applies to any organisation placing AI systems on the EU market or whose AI outputs reach EU users — regardless of where the company is headquartered.

For marketers, the short version: the law doesn't ban AI in marketing. It requires transparency about AI-generated content, prohibits a specific set of manipulative practices, and puts the responsibility for compliance on both the tool vendors and the brands deploying those tools.

Does it affect you?

The Act has a long reach. You're in scope if any of the following apply:

The "targeting EU audiences" clause is the one most non-EU marketers miss. If your Meta or Google campaigns include EU countries in their geo-targeting, the transparency rules apply to that creative.

The four risk tiers — what they mean for marketing

The Act sorts AI systems into four categories. Here's how they map to what marketers actually use:

TierWhat it coversMarketing examples
🚫 ProhibitedSubliminal manipulation, exploiting psychological vulnerabilities, biometric categorisation of people's traits without consentDark-pattern AI that exploits anxiety or urgency below conscious awareness; AI that infers political/religious views from images to target ads
⚠️ High riskAI used in employment, credit, insurance, and profiling of individuals to assess personal characteristicsAI hiring tools; AI credit or insurance scoring; broad profiling systems that feed consequential decisions about individuals
⚠️ Limited riskAI that interacts directly with people or generates synthetic content — must discloseAI chatbots on your site; AI-generated images, video, or copy in ads; deepfake-style synthetic people in creative
✅ Minimal riskEverything else — no mandatory obligationsAI writing assistants, SEO tools, analytics, spam filters, content recommendations, personalisation that doesn't exploit vulnerabilities

The good news: the vast majority of day-to-day marketing AI falls into minimal risk. The action for most teams is in the limited-risk tier — specifically the disclosure requirements for AI-generated content.

Key dates

Enforcement is phased. Here's what's already in effect and what's coming:

February 2, 2025 — Already in force

Prohibited AI practices are banned. The manipulation rules and biometric categorisation restrictions are live now. Penalties for violations: up to €35 million or 7% of global turnover.

August 2, 2025 — Already in force

Obligations for General-Purpose AI (GPAI) model providers began — covering foundation models behind ChatGPT, Gemini, and Claude. Vendors must document training data, publish summaries, and comply with copyright rules.

August 2, 2026 — The big deadline

Full enforcement begins. High-risk AI rules and Article 50 transparency obligations apply. This is the date that matters for most marketing teams: AI-generated content in ads reaching EU audiences must be labelled.

August 2, 2027 — Final phase

The Act applies to all remaining AI systems, including older GPAI models placed on the market before August 2025.

Note: A political agreement reached in May 2026 (the "AI Omnibus") delayed some high-risk AI obligations to 2028 but maintained the August 2026 deadline for transparency and AI-generated content labelling.

Article 50: the disclosure rule every marketer needs to understand

Article 50 is the part that hits marketing teams most directly. From August 2, 2026, it requires:

Importantly, the obligation falls on both the AI tool provider and the brand deploying the tool. If you run an ad with AI-generated creative and don't disclose it, you're liable — even if your ad platform later adds automated labelling. Penalties: up to €15 million or 3% of global turnover.

What the major platforms are doing

The big martech and ad platforms aren't waiting for the deadline:

Don't assume your vendor handles it. Even if a platform labels AI content at the infrastructure level, you — as the deployer — remain responsible for ensuring the label is visible to end users.

5 things marketing teams actually need to do

You don't need an in-house legal team to get this right:

01

Build an AI tool inventory

List every AI tool your team uses that generates or modifies content — copy, images, video, audio, email personalisation. For each tool, note: does it produce output shown to EU users? The inventory is also your starting point for vendor due diligence.

02

Audit for prohibited practices

Review your personalisation and ad targeting logic. AI that adapts messaging based on inferred psychological state, financial stress, or health vulnerability is in the prohibited zone. If your tools use "urgency optimisation" or "emotional targeting," get precise answers from vendors on how these work.

03

Set up an AI content disclosure process

For any AI-generated images, video, or copy running in EU-facing ads: add a disclosure label before August 2026. Agree internally on the format ("Created with AI" is sufficient; elaborate disclaimers are not required). Build it into your creative brief template so it's not an afterthought.

04

Do vendor due diligence

Ask your key AI tool vendors two questions: (1) Are you compliant with the EU AI Act? (2) What watermarking or provenance does your output carry? If they can't answer clearly, that's a risk signal. Check whether they've signed the GPAI Code of Practice.

05

Assign internal ownership

Compliance doesn't need a dedicated role — but it does need a named owner. In most marketing teams this sits with the head of marketing ops or the legal/privacy contact. Someone needs to track the August 2026 deadline and own the vendor review process.

What to watch next

Follow the coverage

📬

AI marketing intel,
delivered weekly.

Join marketers getting the digest free. No spam, always.

No spam · Unsubscribe anytime

EU AI Act for Marketers: What You Actually Need to Know | marketerintel