Sister companyNBM Games: discover our game studio at nbmgames.co
All insightsGovernance

The EU AI Act for business teams: what changes when you deploy AI

A plain-language overview of the risk-based rules, the duty to tell people when they are dealing with an AI, and how it all fits with GDPR and KVKK. General information, not legal advice.

· 6 min read

The EU Artificial Intelligence Act (the AI Act) is the first broad law that regulates AI systems according to what they are used for. It entered into force in August 2024 and applies in stages: the first prohibitions have applied since February 2025, and further obligations phase in over the following years. The EU has also discussed adjusting parts of that timetable, so check the current dates for the obligations that apply to you.

This article is for teams that buy, build or roll out AI tools in a company, not for lawyers. It is general information, not legal advice. For a decision about a specific system, read the official text and guidance, and talk to a qualified lawyer.

Who the Act applies to

The Act distinguishes between roles. A provider develops an AI system, or has one developed, and places it on the market or puts it into service under its own name. A deployer uses an AI system under its authority in a professional context. Most business teams are deployers: they use an AI tool, or a solution built for them. Obligations differ by role, and a deployer that puts its own name on a high-risk system or substantially modifies one can take on provider obligations.

The Act also reaches beyond the EU’s borders. It can apply to a company based elsewhere, for example in Türkiye, when it places AI systems on the EU market or when the output of its AI system is used in the EU.

A risk-based approach

Instead of regulating all AI in the same way, the Act sorts uses into levels of risk. In broad terms:

  • Prohibited practices. A short list of uses is banned outright, such as social scoring, manipulative techniques that cause significant harm, and emotion recognition in the workplace or in education, apart from narrow medical or safety exceptions.
  • High risk. Uses in sensitive areas, including recruitment and other employment decisions, access to essential services such as the credit scoring of individuals, education, critical infrastructure and some uses of biometrics. These carry the heaviest obligations.
  • Transparency obligations. Systems that interact with people or generate content must make that clear, as described below.
  • Minimal risk. Most everyday uses, such as spam filters or tools that draft text for an employee to review, carry no new obligations under the Act.

General-purpose AI models, the large models behind many tools, have rules of their own. Those fall mainly on the companies that develop the models, not on the businesses that use them through a product or an API.

Telling people they are talking to an AI

When people interact directly with an AI system, such as a support chatbot or a voice agent, they must be told that they are dealing with an AI, unless that is obvious from the context. AI-generated or manipulated images, audio or video that could pass as real (deepfakes) must be disclosed. The same applies to AI-generated text published to inform the public on matters of public interest, unless a person has reviewed it and takes editorial responsibility.

In practice this is straightforward to build in: a clear line at the start of a conversation, a label on generated media, and an easy way to reach a person.

Human oversight and documentation

For high-risk systems, deployers must use the system according to the provider’s instructions, assign human oversight to people with the competence, training and authority to carry it out, monitor how the system performs, keep the logs it generates and inform the people affected. Employers must inform workers and their representatives before using a high-risk system at work. Some deployers, such as public bodies and companies that use AI for credit scoring or for life and health insurance pricing, must also assess the impact on fundamental rights.

Even when a system is not high risk, the same habits are worth having. Since February 2025, the Act has also required providers and deployers to take measures so that staff working with AI have sufficient AI literacy: a working understanding of what the tools can and cannot do.

How it relates to GDPR and KVKK

The AI Act does not replace data protection law. When an AI system processes personal data, the GDPR applies in parallel: you still need a legal basis, transparency towards the people concerned and, in many cases, a data protection impact assessment. The GDPR also restricts decisions based solely on automated processing that have legal or similarly significant effects on a person.

For processing in Türkiye, KVKK (Law No. 6698 on the Protection of Personal Data) applies whether or not AI is involved. Among other rights, it lets people object to an outcome against them that results exclusively from automated analysis, and its rules on transferring personal data abroad matter when a model provider processes data outside Türkiye. A company serving both markets will often need to meet the AI Act, the GDPR and KVKK at the same time, and handling them in one process makes that easier.

A practical starting list

  • Make an inventory of the AI systems you use or plan to use, including AI features inside software you already buy.
  • For each one, note your role (provider or deployer) and the likely risk level. Flag anything used for hiring, managing employees, credit, insurance, education or biometrics.
  • Check that every chatbot and voice agent says it is an AI, and that generated media is labelled where required.
  • Name a person responsible for each system, with the authority to pause it.
  • Keep a short record per system: purpose, provider, data used, risk assessment, oversight, and how problems are reported. Ask vendors for their documentation.
  • Train the people who use or oversee AI on its strengths and its limits.
  • Connect all of this to your existing GDPR and KVKK processes, such as records of processing, impact assessments and data processing agreements.

Most of these steps cost little, and they make the rest of the work easier, whatever your final risk classification turns out to be.

Once again: this is not legal advice. The rules and their timelines can change, and the details depend on your system and your role.

Next step

Reading is good. Building is better.

Bring a problem. We will bring a plan.

Or write to us: hello@nbmcore.com