AI is already part of everyday business. Chatbots handle customer enquiries, generative AI produces text, images and video, and AI tools support everything from marketing and journalism to software development and training.
We have previously covered the key obligations under the EU AI Act and their phased introduction (in Finnish), as well as the use of AI in marketing (in Finnish). This article takes a closer look at the Act’s transparency obligations: when do you need to tell users or the public that AI is involved?
The AI Act’s transparency obligations have applied since 2 August 2026. Their main purpose is to help people recognise when they are interacting with AI or encountering content that AI has generated or manipulated. Disclosure is not required for every use of AI, however. The obligations apply to specific situations defined in the Act.
The European Commission published more detailed guidance on applying these obligations in July 2026. The Finnish Transport and Communications Agency, Traficom, has also published practical guidance for organisations in Finland.
When must people be told they are interacting with AI?
As a general rule, if an AI system is designed to interact directly with people, users must be told that they are dealing with AI rather than a human. This covers, for example, AI-powered website chatbots, voice assistants, interactive AI agents and AI avatars.
Users must receive this information no later than their first interaction with the system, unless the use of AI is already obvious in the circumstances. A reference buried in the terms of use, or a technical marker that users cannot see, will generally not be enough. Simply displaying the service provider’s name may not make it clear that the user is talking to AI either.
How must AI-generated content be technically marked?
The AI Act sets out a separate technical marking obligation for providers of generative AI systems. In practice, this applies to companies such as OpenAI and Anthropic when they offer AI systems in the EU that generate text, images, audio or video.
As a rule, providers must ensure that content generated or manipulated by their systems can be identified as such in a machine-readable format. This obligation therefore generally falls on the AI provider, rather than a business using tools such as ChatGPT or Claude.
The technical method can vary depending on the type of content. For supported AI-generated images, OpenAI uses C2PA provenance metadata and an invisible SynthID watermark. An image can be uploaded to OpenAI’s verification tool, which checks the file for signals indicating its OpenAI origin. The watermark itself is invisible to the human eye.
AI-generated text can also carry machine-detectable markings. Anthropic, for example, has described a text watermarking approach for Claude that creates a detectable statistical pattern in the model’s word choices. To a reader, the text looks ordinary, but the provider’s detection method can identify the watermark.
Technical marking is different from a disclosure that users or the public can see. It does not mean that a business using ChatGPT or Claude must add a visible “AI” label to every piece of content.
When do AI-generated images, video or audio need a disclosure?
One of the most practically significant transparency obligations concerns deepfakes. When AI is used in a professional or other non-personal context to generate or manipulate a deepfake, the content must be accompanied by a clear disclosure that it has been artificially generated or manipulated.
A deepfake is AI-generated or manipulated image, audio or video content that resembles existing people, objects, places, entities or events and falsely appears authentic or truthful.
The term therefore covers more than fabricated videos of celebrities. The obligation may also arise in ordinary commercial or creative work. For example, a realistic AI-generated reconstruction in a documentary, or an apparently authentic scene created with AI for an advertisement, may require disclosure. However, not every AI-generated image or video automatically qualifies as a deepfake.
The Act allows some flexibility for creative works. If a deepfake forms part of an evidently artistic, creative, satirical or fictional work, the use of AI-generated or manipulated content must still be disclosed. The disclosure can, however, be made in a way that does not unnecessarily interfere with the presentation of the work. For more on transparency obligations concerning AI-generated audio, particularly in radio broadcasting, see RadioMedia’s blog (in Finnish).
What about AI-written text?
Not every text written with the help of AI needs a visible AI label.
The disclosure obligation applies to AI-generated or manipulated text published to inform the public about matters of public interest. These may include public administration, fundamental rights, public health, environmental protection, and significant economic, political, scientific or cultural developments. The obligation may therefore cover news articles, scientific publications, investor reports and public notices issued by authorities.
Ordinary advertising content, by contrast, generally falls outside this particular obligation.
There is also an important exception: disclosure is not required if the text has undergone human review or editorial control and a person or legal entity holds editorial responsibility for its publication.
Using AI to draft a news article, for example, does not automatically mean that the finished article must carry an AI label, provided the editorial team reviews the content and takes responsibility for publishing it. Proofreading or correcting grammar alone is not enough. The exception requires substantive review or editorial oversight.
What about emotion recognition and biometric categorisation?
Transparency obligations also apply to emotion recognition and biometric categorisation. People exposed to these systems must be informed clearly and prominently, no later than their first exposure.
Emotion recognition may involve, for example, AI analysing a customer’s voice during a customer service interaction to infer whether they are angry or impatient. Biometric categorisation involves assigning a person to a particular group based on biometric characteristics, such as estimating their age group from facial features.
Informing people does not, in itself, make the use of a system lawful. AI systems used to infer emotions in workplaces and educational institutions are generally prohibited, subject to limited exceptions for medical or safety purposes.
Permitted emotion recognition systems and certain biometric categorisation systems also fall within the AI Act’s rules on high-risk AI. In these cases, both the provider and the deployer—the organisation using the system—face extensive obligations beyond transparency.
Who enforces the obligations, and what are the consequences of a breach?
In Finland, compliance with the transparency obligations is supervised by market surveillance authorities. Traficom is generally responsible, although supervision involving high-risk AI systems may fall to the relevant sector-specific authority.
Intentional or negligent breaches of the transparency obligations may result in an administrative fine. For businesses, the maximum fine is €15 million or 3% of total worldwide annual turnover in the preceding financial year, whichever is higher. For small and medium-sized enterprises, the lower of these two limits applies.
These are maximum amounts. Whether a fine is imposed, and its size, is assessed on a case-by-case basis. Under Finnish law, no fine is imposed if the breach is considered minor or if imposing a fine would be manifestly unreasonable.
What should businesses do now?
Start by identifying where customers, users or the wider public encounter AI in your business. Pay particular attention to customer-facing AI systems and externally published content generated or manipulated using AI.
Practical steps include:
Mapping how AI is used in interactions and content that customers or other external audiences encounter.
Giving staff clear guidance on when disclosure is required, how to provide it and who is responsible.
Establishing a clear process for reviewing AI-generated text where needed.
Addressing transparency obligations in contracts with external partners that supply AI solutions or AI-generated content.
The AI Act does not require businesses to disclose every AI-assisted step in their workflow. The key is to recognise when people need to know they are interacting with AI, and when the AI-generated or manipulated nature of content must be explicitly disclosed.
Need help applying the AI Act or assessing how your organisation uses AI? We can help you identify the obligations that apply and put practical processes in place to meet them.
Lila Kallio
Counsel
+358 41 465 1365
Updated: Aug 5
Electronic direct marketing is a key part of sales and customer communications for many companies. However, newsletters, promotional emails and marketing automation only work when there is also a valid legal basis for sending them. Recently, automated marketing calls have attracted particular attention in supervisory practice. In practice, a company should answer three questions before sending a message: who is the recipient, does the message constitute marketing or customer communication, and what is the legal basis for sending it? Electronic direct marketing includes emails, text messages, multimedia messages, voice messages and automated calls used to market products or services to the recipient. The format of the message is not decisive. What matters is its purpose and content. If the objective is to promote the sale of a product or service, the message will generally constitute direct marketing.
A newsletter will usually constitute direct marketing if it contains offers, presentations of services, invitations to events, campaigns or other content intended to promote sales. Even content that appears purely informative may constitute marketing if its main purpose is to encourage the recipient to make a purchase, book a demonstration, download sales materials or attend a commercial event.
However, not every message sent to customers constitutes direct marketing. Order confirmations, delivery notifications, service interruption notices and other messages relating to the performance of a contract or the provision of a service are generally considered customer communications. If sales-oriented content is added to the message, it may nevertheless become direct marketing. For example, adding an offer to “upgrade to the premium version” to a service interruption notice may change how the message is assessed.
Marketing to a personal address generally requires prior consent
As a general rule, electronic direct marketing may only be sent to a natural person if they have given their prior consent. This is the main rule for electronic direct marketing aimed at consumers. The same applies, for example, to automated calling systems. Marketing through automated calls requires the recipient’s prior consent.
Consent must be freely given, specific, informed and unambiguous. In practice, the recipient must take an active step themselves, such as ticking a box to subscribe to a newsletter. A pre-ticked box, silence or the fact that the individual has not specifically objected to marketing does not constitute valid consent.
The requirement that consent be freely given does not necessarily prevent direct marketing consent from being linked to participation in a voluntary prize draw. The Finnish Data Protection Ombudsman has considered that consent to direct marketing requested as a condition for participating in a prize draw may be valid if participation is genuinely voluntary, choosing not to participate causes no detriment and the prize draw is not connected to a service that is essential to the data subject. However, the assessment must always be made on a case-by-case basis.
Separate consent is not always required. If a customer has purchased a product or service from a company and provided their electronic contact details in connection with the purchase, the company may, subject to certain conditions, market its own similar products or services to that customer.
However, this exception cannot be treated as a general authorisation for all marketing. It requires that the contact details were obtained directly from the customer in connection with a sale, that the marketing concerns the same company’s own products or services, and that the products or services being marketed are similar to the customer’s previous purchase. For example, if a customer has purchased a particular software service, it may, depending on the circumstances, be permissible to market additional features within the same service package or other similar services to them. The same contact details do not, however, permit the marketing of services offered by an entirely different group company, products offered by a business partner or products belonging to a completely different category.
The customer must also be given an easy and free opportunity to object to marketing both when their contact details are collected and in every subsequent marketing message.
In B2B marketing, it is not enough that the recipient works for a company
In business-to-business marketing, the assessment depends on the type of address to which the message is sent and the recipient’s role. As a general rule, electronic direct marketing may be sent without prior consent to general company addresses, such as info@company.com or sales@company.com. However, the message must still provide an opportunity to object to further marketing.
If the message is sent to a named individual, such as firstname.lastname@company.com, the sender must assess whether the product or service being marketed is materially related to the recipient’s professional duties. For example, marketing financial administration services to a person responsible for financial decision-making may be justified. The same message cannot, however, be sent as a precaution to every individual whose contact details can be found at the company.
The essential point is that the recipient’s role and the content being marketed must correspond with one another. A mere assumption that the recipient might be interested is not enough. In B2B marketing, it is also important to remember that a personal work email address does not become a freely available marketing channel simply because the address is publicly available or can be found on LinkedIn, a website or an event participant list. The sender must still be able to explain why the message sent to that particular person relates to their professional duties and why the marketing has been appropriately targeted.
Companies must be able to demonstrate consent
A company must be able to demonstrate the basis on which it sends electronic direct marketing. If the marketing is based on consent, the documentation should indicate at least when the consent was given, through which channel it was given, the type of marketing to which it applies and the information provided to the recipient when consent was requested.
Every electronic direct marketing message must include an easy and free method of objecting to the marketing or withdrawing consent, such as a functioning unsubscribe link. Withdrawing consent must be as easy as giving it. In practice, recipients should not be required to contact customer service separately or log in to a complicated system merely to unsubscribe from a newsletter.
Electronic direct marketing usually involves the processing of personal data, which means that the requirements of the General Data Protection Regulation also apply. The recipient must therefore be clearly informed about how their personal data is processed.
In practice, a company should ensure that its privacy notice explains who processes the personal data, the purposes for which the data is used, the legal basis for the processing, the source of the data, how long the data is retained and how the recipient can exercise their rights.
The regulation of electronic direct marketing does not prevent effective marketing, but it does require companies to handle the fundamentals carefully. Before launching a campaign, a company should verify where the recipients’ contact details came from, the legal basis for sending the messages and how marketing objections and withdrawals are implemented in practice.
Lila Kallio
Counsel
+358 41 465 1365
To receive our articles directly in your inbox, subscribe to the Folks newsletter here.
AI tools are developing at a remarkable pace. Not long ago, an AI-generated image was easy to recognise, but today, the best AI-generated images are so convincing that they can be indistinguishable from genuine photographs. AI can also be used to produce text, audio and video, offering creative industry professionals significant opportunities to accelerate content production and develop new forms of expression.
At the same time, the use of AI raises numerous legal questions. In this blog post, I discuss the legal considerations that companies operating in the creative industries should take into account before introducing AI tools and using them as part of their creative work.
Who owns AI-assisted content, and how can it be protected?
A key question is whether the creator obtains an exclusive right to use and license material produced with the assistance of AI, or whether the result remains freely available for anyone to use. There is no straightforward answer, as it depends on the extent of the human creator’s own creative contribution to the work.
Copyright belongs to the person who creates a work, provided that the work is sufficiently independent and original. Copyright protection always requires a creative contribution by a human, and the assessment is made on a case-by-case basis. Copyright never protects an idea as such, but only the specific form in which the idea is expressed.
When a creative professional uses AI as a tool in their work, the existence of copyright is assessed according to these same principles. Copyright does not arise where a person gives the AI only a general instruction and the AI produces the final content entirely without any creative contribution from the user.
The situation is different if AI is used as part of a broader creative process. If the creator uses AI, for example, to support brainstorming, then creates the content themselves and uses AI only for final refinements, the work is likely to contain enough of the creator’s own creative contribution to qualify for copyright protection.
The use of AI should also be documented in case of potential disputes. Retaining the creator’s own drafts and the prompts used can make it easier to demonstrate which parts of the final result are based on human creative work.
If content is created using generative AI and the result does not qualify for copyright protection, protection may in some cases also be sought through trademarks. For example, Moomin Characters has protected Moomin characters as trademarks. However, it is important to note that a trademark only provides protection in specified classes of goods and services and within a particular geographical area. It does not protect the creative content as such. Its scope and purpose therefore differ from those of copyright protection.
If a creative project involving extensive use of generative AI is being planned, trademark protection may nevertheless be worth considering alongside copyright as a complementary form of protection.
Can AI-generated material be used freely?
Material created with AI is not automatically free from third-party rights. The key questions are which AI service was used to produce the material and how liability has been allocated in the service’s terms of use.
Some AI services use only licensed training data or other material that is not protected by copyright and contractually assume responsibility for ensuring that the materials generated by the AI do not infringe third-party copyrights. In such cases, the service provider bears responsibility for the output to the extent agreed in the terms of use, and the user’s legal risk is substantially lower.
Many widely used services, however, have been trained on extensive datasets collected from the internet, which may also contain copyright-protected material. In most services, the service provider does not contractually assume responsibility for ensuring that the generated content does not infringe third-party rights. Instead, responsibility for using the content remains with the user. Material produced using such services therefore requires careful review before publication.
In summary, there is no universally applicable answer as to whether AI-generated materials may be used freely. The allocation of responsibility depends primarily on the service used and its terms of use, which should be reviewed before AI is introduced as part of content production.
Can confidential information be entered into an AI service?
As a general rule, confidential information should not be entered into an AI service unless the service is specifically intended for business use and its terms and information security have been carefully assessed.
For business purposes, companies should use business or enterprise versions of generative AI services unless they operate a local AI solution on their own servers. In consumer services, such as the free and Plus versions of ChatGPT, materials and prompts entered into the service may, in accordance with the terms of use, be used to train the AI model. In addition, the service provider may reserve broad rights to use and even share material uploaded by users for purposes other than AI model training.
This means that if a company’s employees use consumer licences in their work, they may compromise the confidentiality of information by entering confidential material into the service.
The practical risk may arise, for example, where a screenwriter or copywriter uses a free consumer AI service to refine a text and enters an unpublished script or campaign concept into the service. Under consumer licences, the content may be used to develop AI models, meaning that the material is no longer under the company’s exclusive control. Even if the content does not appear elsewhere in an identical form, there is a risk that recognisable features of the material may be used as part of outputs generated for other users.
It is also important to note that other risks do not disappear even if the use of content for training purposes can be prohibited separately in the service settings or terms of use. Under consumer licences, service providers often do not provide binding guarantees regarding the security of the service. Information entered into the service may therefore be exposed to data breaches or other information security incidents. Protecting confidential information requires selecting a licence suitable for business use and assessing the service’s terms and information security before introducing the AI service.
Can personal data be entered into an AI service?
If personal data, such as a person’s image, voice or name, is entered into an AI service, the General Data Protection Regulation, or GDPR, applies.
The company is responsible for ensuring that the licensing terms of the AI service have been carefully assessed. As part of its compliance with the GDPR, the company must ensure, among other things, that material uploaded to the service is not used to train the AI, that it is possible to enter into the data processing agreement required by the GDPR with the service provider, and that the AI service provides appropriate security for the protection of personal data. In practice, the use of an AI service will generally also require a data protection impact assessment.
The use of AI services under free consumer licences will generally not comply with the GDPR, as service providers often also use uploaded material for their own purposes.
Is it permissible to use AI to generate images or voices of real people?
Particular attention is also required where AI is used to produce images of real people or to create AI-generated copies of a real person’s voice. A person’s image, voice and other identifying features constitute personal data. AI-generated material may also be considered personal data if a specific individual can be identified from it. In such cases, all GDPR requirements apply, including the requirement to inform the individual about the processing of their personal data and the requirement to have a lawful basis for the processing.
In content production, it is also important to recognise that using a person’s image or voice for commercial purposes requires that person’s consent. In 2025, the Helsinki Court of Appeal ordered an underwear company to pay Jasper Pääkkönen EUR 300,000 in compensation for the unauthorised use of his name, image and voice in an extensive advertising campaign. The judgment is not yet final, and the Supreme Court has granted leave to appeal. The obligation to obtain consent also applies where the person’s image or voice has been generated using AI.
When must AI-generated content be labelled as a deepfake?
The EU’s new AI Act introduces transparency requirements concerning deepfakes. The Act will become applicable gradually, and the provisions concerning deepfakes will apply from 2 August 2026.
Deepfakes are AI-generated or AI-manipulated image, audio or video content that resembles existing people, objects, places, entities or events and may falsely appear to the recipient to be authentic or truthful. The transparency obligation is therefore not limited to deepfakes depicting people.
Under the AI Act, deepfakes must be clearly labelled as having been artificially generated or manipulated. However, the Act includes an exception for creative works. Where the content forms part of an evidently artistic, creative, satirical, fictional or similar work or programme, the disclosure may be made in a manner that does not interfere with the display or enjoyment of the work.
For example, a documentary-style television production may use highly realistic AI-generated images or videos to illustrate historical events. If the context does not otherwise indicate that the material is artificial, viewers may believe it to be genuine archival footage. In such a case, the transparency obligation under the AI Act must generally be considered. However, the exception for creative works allows the use of AI to be disclosed in a manner appropriate to the nature of the work, for example in the programme’s end credits, provided that the audience is not left with a false impression of the authenticity of the content.
A breach of the transparency obligation concerning deepfakes may result in an administrative fine under the AI Act.
Checklist for creative industry professionals
Rights: ensure that your own creative contribution is sufficient if you want the final result to qualify for copyright protection.
Documentation: document the use of AI and your own creative contribution.
Terms of use: review the service terms and determine what rights you receive to the generated content and who is responsible for potential infringements.
Business-level licences: only use business-level licences for professional purposes and enter into a data processing agreement where necessary.
GDPR and consent: using a person’s image or voice requires compliance with data protection rules and, for commercial use, the person’s consent.
Deepfakes: if content has been artificially generated or manipulated, ensure that the disclosure requirements of the AI Act are met.
Would you like to discuss the legal questions surrounding AI? We help companies assess the legal risks and opportunities associated with the use of AI in the creative industries and in other AI-enabled business operations.
Lila Kallio Counsel
+358 41 465 1365
Katri Aarnio Counsel
+358 50 306 2031
To receive our articles directly in your inbox, subscribe to the Folks newsletter here.