Every time you scroll through social media, watch a video, or read an article online, you may soon see a small label: "AI-generated." That label is about to become mandatory across the European Union, and the bloc has just published the rulebook on how to do it right.
What the EU's AI content labelling playbook actually requires
The European Commission released the final Code of Practice on June 10, 2026, setting out practical steps for businesses that build and use generative AI to mark and label what their systems produce. The code is voluntary, but the obligations it points to are not.
From August 2, 2026, two things must be clearly flagged. Deepfakes — AI-generated or manipulated audio, video, or images that appear realistic — must be labelled. Any AI-generated text, if published in a context where people might be misled, must also carry a disclosure.
Why this matters for every Indian business using AI
If your company uses AI tools to generate product images, marketing copy, customer service responses, or even internal reports, and you operate in or serve the EU market, these rules apply to you. The EU AI Act has extraterritorial reach — any company that places AI systems on the EU market or whose AI outputs affect EU citizens must comply.
For Indian IT firms, SaaS startups, and e-commerce platforms with European customers, this is not a distant regulation. It is a compliance deadline that is 53 days away from the code's publication.
How the Code of Practice works: voluntary but strategic
The Code itself is optional. Signing it gives a business a recognised way to show it complies with Article 50. Companies that do not sign must still meet the same transparency obligations — they just have to prove compliance through other means.
The European Commission's Digital Strategy unit published the code, which includes technical specifications for watermarking, metadata tagging, and disclosure formats. The goal is to create a standardised approach across the bloc, reducing confusion for businesses and making it easier for users to identify AI-generated content.
Who is affected: from tech giants to small startups
The rules apply to both providers — companies that build generative AI systems — and deployers — businesses that use those systems to create content. A small marketing agency using ChatGPT to write ad copy is as responsible as OpenAI itself, if the output reaches EU users.
This creates a compliance burden across the supply chain. Providers must build labelling capabilities into their systems. Deployers must ensure those labels are visible and accurate in the final output.
European Commission's stance: transparency as a trust-building tool
Officials in Brussels have framed the labelling rules as a way to protect the integrity of the information ecosystem. The risks of deception and manipulation from AI-generated content are well-documented — from fake news to deepfake fraud. The Commission believes mandatory labelling will help users make informed decisions about what they see and hear.
"These transparency obligations address risks of deception and manipulation, fostering the integrity of the information ecosystem," the Commission stated in its policy document on the Code of Practice.
What Article 50 of the EU AI Act actually says
Article 50 is the legal backbone of the labelling requirements. It mandates that providers of generative AI systems ensure their outputs are "marked in a machine-readable format and detectable as AI-generated." Deployers must also inform users when they are interacting with AI-generated content, unless it is obvious from the context.
The article covers text, images, audio, and video. Deepfakes receive special attention because of their potential to mislead people about the authenticity of what they are seeing or hearing.
Confirmed facts vs what remains unclear
Confirmed: The Code of Practice was published on June 10, 2026. Article 50 obligations become mandatory on August 2, 2026. The code is voluntary but the underlying law is not. Deepfakes and AI-generated content must be labelled.
Unclear: The exact technical standards for watermarking and metadata are still being finalised in some areas. The Commission has not yet specified penalties for non-compliance, though the EU AI Act framework allows for fines up to 7% of global annual turnover for the most serious violations. It is also unclear how enforcement will work for non-EU companies that do not have a physical presence in the bloc.
Risks and challenges for businesses
Compliance costs are a major concern. Small and medium businesses may struggle to implement labelling systems, especially if they use multiple AI tools from different providers. There is also the risk of "label fatigue" — if every piece of content carries an AI label, users may stop paying attention.
Critics argue that the rules are too broad. A simple AI-assisted grammar check on a blog post could technically require a label, though the Commission has indicated that de minimis uses may be exempt. The line between AI-generated and human-created content is increasingly blurry, and the rules may struggle to keep pace with technology.
Wider trend: global push for AI transparency
The EU is not alone in demanding AI content labels. China has already implemented mandatory labelling for AI-generated content. The United States is developing its own framework through executive orders and voluntary commitments. India's Ministry of Electronics and IT has also signalled interest in AI transparency rules.
The EU's approach is the most comprehensive and legally binding so far, and it is likely to influence regulations in other jurisdictions. Companies that comply with the EU rules will be well-positioned for similar requirements elsewhere.
What businesses should do now
If your company uses generative AI and serves EU customers, start auditing your AI outputs immediately. Identify which content needs labelling. Work with your AI providers to ensure their systems support the required metadata and watermarking. Consider signing the Code of Practice to simplify compliance.
For Indian companies, this is also a moment to review your data protection and AI governance frameworks. The EU AI Act overlaps with GDPR in several areas, and non-compliance with one could trigger scrutiny under the other.
Future outlook: what happens after August 2
Once the rules take effect, enforcement will begin. The European Commission has not yet announced specific inspection or penalty procedures, but national regulators in each EU member state will be responsible for oversight. Companies found non-compliant could face fines, orders to cease operations, or reputational damage.
The Code of Practice is expected to evolve as technology changes. The Commission has indicated it will update the code periodically to address new forms of AI-generated content and emerging risks.
Our Take
The EU's AI content labelling playbook is a landmark step in the global regulation of artificial intelligence. It moves the conversation from voluntary pledges to enforceable law. For businesses, the message is clear: transparency is no longer optional. The August 2 deadline is tight, but the cost of non-compliance is far higher than the cost of preparation. This is not just about avoiding fines — it is about building trust with users in an era where seeing is no longer believing.
Frequently Asked Questions
What is the EU AI content labelling playbook?
It is a voluntary Code of Practice published by the European Commission on June 10, 2026, that provides practical guidance for companies on how to label AI-generated content to comply with Article 50 of the EU AI Act.
When do the AI content labelling rules become mandatory?
The transparency obligations under Article 50 of the EU AI Act become mandatory on August 2, 2026. Companies must label AI-generated content from that date, whether or not they sign the Code of Practice.
Do these rules apply to Indian companies?
Yes, if your company places AI systems on the EU market or if your AI-generated content reaches EU citizens. The EU AI Act has extraterritorial reach, similar to GDPR.
What happens if a company does not comply?
Non-compliance can lead to fines under the EU AI Act framework, which allows penalties up to 7% of global annual turnover for the most serious violations. National regulators in each EU member state will enforce the rules.
What types of content need to be labelled?
Deepfakes (AI-generated or manipulated audio, video, or images that appear realistic) must be labelled. AI-generated text must also be disclosed if it could mislead users about its origin.
