Issue #22: The EU just made it illegal to publish AI content without disclosing it. Here's what small businesses need to do.

August 2, 2026. A new law kicks in across the European Union. Fines up to €15 million. And it applies to businesses outside the EU too — if your content reaches EU users.

If you use AI to create content, run a chatbot, or communicate with customers in any EU country — the law changed for you last week. Most small business owners have no idea. This issue gives you the plain-English version and a practical checklist to get compliant without hiring a lawyer.

Article 50 is the transparency chapter of the EU's Artificial Intelligence Act — a regulation that came into force in August 2024 with a two-year phased rollout. August 2, 2026 is the date that Article 50 became fully enforceable. The European Commission published the implementing guidelines on July 20, just 13 days before the deadline.

In plain English, Article 50 says: if you're using AI in ways that people might mistake for human, you have to tell them.

It covers four specific situations:

1. AI chatbots and virtual assistants — if you run an AI chatbot on your website, a WhatsApp bot, an AI customer service agent, or any AI system that talks directly with customers — you must clearly disclose that they're interacting with AI, not a human. This applies unless it's already obvious (like a clearly-labelled robot icon). The disclosure must happen at the point of interaction, before the conversation starts.

2. AI-generated content (images, audio, video, text) — if you produce synthetic content using AI and publish it publicly, it must be marked in a machine-readable format that allows detection tools to identify it as AI-generated. This is a technical requirement, not just a visible label — though a visible label is also recommended.

3. Deepfakes and manipulated media — if you create or publish video or audio that uses AI to make a real person appear to say or do something they didn't — you must label it clearly as artificially generated or manipulated. HeyGen videos using your own digital twin are not deepfakes. AI-generated content featuring real third-party people is.

4. AI-generated news and public-interest content — text generated by AI and published as news, opinion, or public-interest content must be labelled. This catches AI-written blog posts, newsletters, and social media content published on topics of public concern.

Who it applies to: Any business, anywhere in the world, whose AI content or AI systems are used within the EU. You don't need to be based in Europe. If EU users see your content or interact with your chatbot, you're in scope.

The verdict: Most small businesses using AI for customer communication and content production have obligations under Article 50 they haven't addressed yet. The compliance steps are not technically complex. What they require is awareness — which is exactly what most small business owners don't have right now.

A US-based SaaS company selling project management software had EU customers representing about 30% of their user base. They ran an AI chatbot for customer support, published a weekly newsletter with AI-assisted content, and used HeyGen videos for product explainers.

Their Article 50 exposure, mapped out in 30 minutes: the chatbot needed a clear "You're chatting with an AI" disclosure before conversations started. Their newsletter needed a brief note about AI assistance in the writing process. Their HeyGen videos — using their own avatar with their own voice — were not in scope for deepfake rules but fell under the AI-generated content marking requirement.

They fixed all three in one afternoon: updated the chatbot welcome message, added a two-line disclosure to the newsletter footer, and added an "AI-assisted" tag to video descriptions. No legal team required. Total time: four hours. Total cost: zero.

Work through each item and check off what applies to your business:

Step 1: Map your EU exposure Do you have any EU customers, website visitors, email subscribers, or social media followers? If yes, you're in scope. If you're genuinely certain your content and services reach zero EU users, you can stop here.

Step 2: Audit your AI chatbots and customer-facing AI List every AI system that interacts with customers — website chat, WhatsApp automation, email auto-responders, social media DM bots. For each one, add a clear disclosure at the first point of contact: "You're chatting with an AI assistant." This can be as simple as a first message, a labelled chat widget, or an "AI" badge on the interface.

Step 3: Audit your AI-generated content List everything you publish that was created or substantially assisted by AI — blog posts, social media content, newsletters, marketing copy, videos. For each type, decide how you'll disclose: a footer note ("Created with AI assistance"), a visible label, or a post caption. Visible human labels count for Article 50 purposes alongside the machine-readable requirement.

Step 4: Check your video content HeyGen videos using your own digital twin and voice: add an "AI-generated" disclosure in the video description or caption. AI-generated images or audio featuring real third-party people: these are in deepfake territory — don't publish without explicit consent and clear labelling.

Step 5: Update your privacy policy or terms Add a brief section noting that you use AI systems in customer-facing roles. This is both a legal best practice and increasingly expected by customers.

Step 6: Document what you've done Keep a brief record of the changes you made and when. If enforcement ever comes knocking, demonstrating that you took compliance seriously is meaningful mitigation. A simple Google Doc with dates is fine.

"I'm a small business in the US/Australia. Why does an EU law affect me?"

Because the EU AI Act applies based on where your content is used, not where you're based. This is the same principle as GDPR — which caught global businesses by surprise for the same reason. If you have EU visitors to your website, EU subscribers to your newsletter, or EU users of your chatbot — you're using AI in the EU from the regulator's perspective. The fine structure scales with company size: up to €15 million or 3% of worldwide annual turnover, whichever is higher. For most small businesses, the realistic enforcement risk in the short term is low — regulators will focus on large-scale, egregious violations first. But "low enforcement risk now" is not the same as "not my problem." The law is in force, the compliance steps are simple, and getting ahead of it takes an afternoon.

The compliance steps are genuinely low-cost and low-effort for most small businesses. The gap between "I use AI" and "I'm Article 50 compliant" is a few text changes and a policy update.

This week: spend 30 minutes mapping your EU exposure and your AI usage. Two lists: (1) which of your customers or audience members are in the EU, and (2) which AI tools you use in customer-facing roles or content production. That map tells you exactly what you need to disclose and where. Most small businesses will find the compliance is three to five text changes, not a legal project.

The EU AI Act is the most comprehensive AI regulation in the world, and its phased rollout means more obligations kick in over time. The prohibited practices chapter came into force in February 2026. Article 50 transparency requirements came into force August 2, 2026. High-risk AI system requirements for new systems also apply from August 2, 2026. General-purpose AI model obligations have been in force since August 2025. The EU is building the compliance infrastructure incrementally — which means businesses that start now have a significant advantage over those who wait for enforcement to force the issue.

Panic-buying legal services to audit your entire AI stack for EU AI Act compliance if you're a small business with basic, standard AI usage. For most small business owners using ChatGPT for content, a chatbot for customer support, and HeyGen for video — the compliance steps are simple, free, and practical. A full legal audit is appropriate for businesses with high-risk AI systems, complex data processing, or significant EU revenue. For everyone else: read the checklist above, make the changes, document what you did, and move on. Don't let the €15 million headline send you into disproportionate response mode.

Until next issue, Kris

The Layman's AI — The only AI updates your business actually needs.