Using AI in Marketing? International Rules May Require You to Tell Customers

Artificial intelligence has become part of everyday marketing faster than most companies have developed rules for using it. Marketing teams are relying on AI to draft social posts, create images, generate video, personalize advertising, answer customer questions, produce voiceovers, edit photographs, and develop campaign concepts. For many businesses, those tools have become so routine that the legal question is no longer whether AI is being used, but whether anyone inside the company knows when that use needs to be disclosed.

That question has become more important for U.S. companies doing business in Europe. Since August 2, 2026, transparency obligations under Article 50 of the European Union's AI Act have been in effect. Those rules address several uses that can overlap directly with marketing and customer communications, including AI systems that interact with people, synthetic images and video, deepfakes, AI-generated text on matters of public interest, and certain AI-generated content that must be technically identifiable as artificial. Digital Strategy

For a Chicago retailer selling to customers in France, an Illinois technology company operating an AI chatbot for European users, or a U.S. brand running a digital advertising campaign across the EU, this is not simply a European technology-policy story. It is becoming part of the practical question of how businesses communicate with customers when AI is involved.

Using AI Does Not Automatically Mean Every Piece of Content Needs a Disclaimer

The first thing businesses should understand is that the EU AI Act does not create a blanket rule requiring every AI-assisted email, advertisement, caption, or photograph to be labelled “created by AI.” The requirements depend on the type of system, what the AI actually did, how the resulting material is used, and whether the business is acting as the provider of the system or as the organization deploying it.

That distinction matters because AI can play very different roles in a marketing department. An employee might use an AI tool to reorganize a paragraph written by a human, correct grammar, resize an image, or remove background noise from a video. At the other end of the spectrum, a company might generate a realistic spokesperson who does not exist, create synthetic footage of a real person appearing to say something they never said, operate an AI customer-service representative, or publish public-facing information written substantially by an AI system.

Those situations are not treated the same way.

Article 50 specifically provides an exception from certain technical marking requirements where an AI system merely performs a standard editing function or does not substantially alter the input or its meaning. At the same time, the regulation places clearer disclosure obligations on deepfakes and certain other types of synthetic content. Eur-Lex

For marketers, the practical lesson is that the right internal question is not simply, “Was AI involved?” It is “What did the AI create or change, and how are we presenting the result to the audience?”

Customers May Need to Know When They Are Interacting With AI

Customer-service automation is one of the clearest examples.

Businesses increasingly use AI-powered chatbots and virtual agents to answer product questions, recommend products, handle basic complaints, book appointments, qualify leads, and guide customers through purchases. Some of these systems are obviously automated. Others are intentionally designed to communicate in a highly conversational way that may make the interaction feel human.

Under Article 50, providers of AI systems intended to interact directly with people must generally ensure that users are informed that they are interacting with an AI system unless that fact would already be obvious to a reasonably well-informed and observant person in the circumstances. The disclosure must be provided clearly and no later than the person's first interaction or exposure. Eur-Lex

For marketing and customer-experience teams, that means the design of an AI chatbot is also a compliance issue. If a company presents an automated agent in a way that could reasonably lead customers to believe they are communicating with a person, the business should not assume that the technology can remain invisible simply because a third-party vendor supplied it.

A U.S. company can also come within the AI Act even without a European headquarters. The regulation reaches certain providers outside the EU that place AI systems on the EU market and certain providers or deployers outside the Union where the output of the AI system is used in the EU. Eur-Lex

That makes customer-facing AI worth reviewing whenever a U.S. business is actively serving European customers.

Deepfake Advertising Creates a Much Clearer Disclosure Problem

The rise of synthetic video and audio is creating a different set of marketing risks.

Generative tools can now produce highly realistic people, voices, locations, events, testimonials, and product demonstrations. A company can create a virtual spokesperson without hiring an actor, place a product into an environment that was never photographed, or generate a voice that sounds remarkably human. The technology can be useful, but the more realistic the content becomes, the more important transparency can become.

Under the EU AI Act, deployers of AI systems that generate or manipulate image, audio, or video content constituting a deepfake must disclose that the content has been artificially generated or manipulated. The regulation describes deepfakes broadly in relation to synthetic or manipulated content that resembles real people, objects, places, entities, or events and could falsely appear authentic or truthful. Eur-Lex

For marketers, this could be relevant to far more than political misinformation. Imagine a company using AI to create a realistic video of a well-known executive appearing to introduce a product when the executive never actually recorded the statement. Consider an advertisement showing a real event, store, destination, or customer experience that never occurred. Even a campaign involving an AI-altered real person can raise different questions from a clearly stylized illustration or obviously fictional character.

The Act does contain more flexible treatment for content that is evidently artistic, creative, satirical, fictional, or similar. In those situations, the disclosure requirement is intended to be handled in a way that does not interfere unnecessarily with the experience of the work. Eur-Lex

The key issue is whether the audience could reasonably mistake synthetic material for authentic content.

AI-Generated Images, Video, Audio, and Text May Need Technical Marking

The Act also places responsibilities on the companies providing generative AI systems.

Providers of AI systems capable of generating synthetic audio, images, video, or text are generally required to ensure that their outputs are marked in a machine-readable format and can be detected as artificially generated or manipulated. The technical solution is expected to be effective, interoperable, robust, and reliable to the extent technically feasible. Eur-Lex

This is different from a visible disclosure to a consumer. A machine-readable marker is designed to help platforms, regulators, and other systems detect the artificial origin of content.

That distinction matters to marketing departments because businesses frequently use third-party AI platforms rather than developing their own models. A company may therefore depend on its technology vendor to satisfy technical marking requirements while remaining responsible for how the content is actually deployed or disclosed to consumers.

The European Commission's guidance on Article 50 was published specifically to help providers and deployers distinguish between these roles and apply the transparency requirements consistently. Digital Strategy

Businesses purchasing generative AI tools should therefore understand what their vendors provide. If a platform claims that generated images, audio, or video include compliant machine-readable markings, the business should know what that statement means and whether the feature remains intact when content is edited, downloaded, republished, or moved through another platform.

AI-Written Marketing Copy Is More Nuanced

Text presents a more complicated issue because the EU AI Act does not require every piece of AI-generated commercial copy to carry a visible label simply because generative AI helped write it.

Article 50 contains a specific disclosure rule for AI-generated or manipulated text when that text is published for the purpose of informing the public on matters of public interest. Even then, the Act provides an exception where the content has undergone human review or editorial control and a natural or legal person holds editorial responsibility for the publication. Eur-Lex

That means an AI-assisted product description, promotional email, or ordinary advertising slogan does not automatically fall into the same category as AI-generated public-interest reporting.

Businesses should still be careful about how AI-generated claims are used. Traditional advertising, consumer-protection, intellectual-property, privacy, and sector-specific laws continue to apply regardless of whether AI produced the underlying copy. AI does not provide a defense for a false product claim, unauthorized use of someone else's intellectual property, or misleading advertising.

The regulatory question may therefore differ from the ordinary marketing-law question. A particular advertisement might not require an AI disclosure under Article 50 while still raising legal issues because of what the advertisement says or depicts.

For businesses, this is another reason not to build an AI policy around one rule alone.

AI Influencers and Virtual Spokespeople Deserve Particular Attention

Virtual influencers, synthetic presenters, and AI-generated brand personalities are increasingly common in advertising. These tools can be useful because companies can create consistent characters across campaigns without traditional production constraints, but they also illustrate why AI transparency is becoming a marketing issue rather than merely a technology issue.

A clearly fictional animated mascot is unlikely to create the same audience expectations as a photorealistic synthetic person presented as though they were a real customer, employee, expert, or influencer. The more a campaign relies on realism, the more important it becomes to consider whether viewers could reasonably misunderstand what they are seeing.

That analysis becomes especially important when an AI persona appears to provide a testimonial. If a synthetic individual describes personal experience with a product, consumers may understand that statement differently from an ordinary fictional advertisement. Businesses should therefore consider not only AI-specific disclosure requirements but also existing advertising rules governing endorsements and misleading representations.

The EU AI Act adds another layer to that analysis where the material qualifies as a deepfake or other covered synthetic content. Eur-Lex

The practical point is that a marketing team should decide whether a campaign needs disclosure before the content is published, not after someone asks whether the person in the advertisement is real.

Agencies and Contractors Do Not Make the Issue Disappear

Many businesses outsource creative work to advertising agencies, social-media firms, freelancers, public-relations consultants, and content studios. Those providers increasingly use generative AI as part of their ordinary workflows, sometimes without discussing every individual tool with the client.

That can create a visibility problem.

A company may approve an advertisement without realizing that an agency generated the spokesperson, cloned audio, substantially altered a photograph, or produced most of the underlying content using AI. If the campaign is then deployed to European consumers, questions may arise about which party is responsible for disclosures and whether the appropriate technical or visible labels were preserved.

Contracts with marketing vendors should therefore address AI use where it is material to the work. A company may want to know when generative AI is being used to create customer-facing content, whether third-party intellectual property is involved, who is responsible for required disclosures, and whether the vendor will preserve technical provenance or labelling features when necessary.

This does not require a lengthy AI clause in every routine freelancer agreement. The level of contract protection should reflect the type of campaign and the risk involved. But businesses should not assume that outsourcing the creative work also outsources legal responsibility for how the finished advertising is used.

Marketing Teams Need Rules They Can Actually Use

A workable AI marketing policy should not require employees to consult legal counsel every time they use a writing assistant.

Instead, the business should identify the kinds of activity that warrant additional review. Routine editing may sit at one end of the spectrum. Customer-facing AI agents, realistic synthetic people, voice cloning, deepfake-style video, AI-created testimonials, biometric technology, and public-interest content may sit much closer to the other.

Employees should know when disclosure may be required, when a campaign needs legal review, what types of AI tools are approved, and whether there are restrictions on entering confidential or customer information into third-party systems.

The business should also know which markets the content will reach. A social campaign created in Chicago can be viewed immediately in Paris, Madrid, or Berlin. A chatbot embedded on a global website may serve European consumers even if the company originally developed it for U.S. customers.

Digital marketing has made geographic boundaries less obvious. Regulation has not disappeared simply because the campaign crosses those boundaries automatically.

The Larger Issue Is Trust, Not Just Compliance

Transparency rules are partly a response to a broader problem: generative AI is making it increasingly difficult for customers to know what is authentic.

That creates a business issue as well as a regulatory one.

If customers discover that a testimonial was synthetic, a spokesperson was not real, a product demonstration was fabricated, or they were communicating with a chatbot they believed was a person, the resulting loss of trust may matter more commercially than the disclosure itself.

For that reason, businesses should not approach AI transparency only as a question of what they are legally required to reveal. There is also value in determining what the customer would reasonably want to know.

In some campaigns, clearly identifying AI involvement may strengthen rather than weaken the creative concept. In others, disclosure can prevent the audience from drawing conclusions the company never intended.

The most sustainable use of AI in marketing is likely to be one in which the technology improves communication without depending on customers being confused about what they are seeing.

What Businesses Should Review Now

Companies marketing to European audiences should review where AI is already appearing in their customer communications. That includes chatbots, automated sales agents, synthetic images, video and audio, virtual influencers, AI-generated testimonials, public-facing text, personalized advertising, and content created by outside agencies.

The review should distinguish between ordinary AI assistance and content that could create a meaningful transparency obligation. Businesses should also identify which vendors provide their AI tools, what those vendors do to mark or identify synthetic content, and whether downstream editing or publishing removes those features.

Marketing, legal, technology, and customer-experience teams should agree on when disclosure is required and when a campaign needs additional review. That does not need to become an elaborate approval process for every use of AI. The objective is to prevent higher-risk content from being published simply because no one inside the company realized that different rules applied.

The Bottom Line

AI can make marketing faster, less expensive, more personalized, and more creative. None of those advantages are likely to disappear. What is changing is the assumption that businesses can always use artificial content without telling the audience that AI was involved.

The EU AI Act's transparency requirements are one of the clearest examples of that shift. Customer-facing AI may need to identify itself. Deepfakes may need visible disclosure. Generative systems may need machine-readable markings. Certain public-interest text may require disclosure depending on how it was created and reviewed. At the same time, routine editing and many ordinary uses of AI do not automatically require the same treatment. Eur-Lex

For U.S. companies operating internationally, the sensible approach is not to label every piece of AI-assisted marketing. It is to understand which uses create genuine transparency obligations and build those requirements into the marketing process before a campaign reaches customers.

This article is provided for informational purposes only and does not constitute legal advice.

Using AI in advertising, customer service, social media, or other marketing that reaches customers in Europe? TEIL Firms can help you determine which AI transparency rules may apply, review customer-facing disclosures and vendor agreements, and build practical compliance protections around your marketing activities. Click the button below to get legal support with your international AI and marketing compliance.