EU AI Act
The EU AI Act for companies that use AI.
Most companies do not develop AI, they use it. The AI Act calls them deployers. This page gives a first overview of the topics that matter for deployers and where a platform can help.
As of October 2026
Provider or deployer.
Which requirements apply depends on the role. A company that uses AI in-house is usually a deployer. In each case it comes down to the specific use.
Providers
Develop an AI system or place it on the market under their own name, such as the makers of models and AI platforms.
Deployers
Use an AI system under their own authority, such as a company that gives its employees an AI chat. Anyone who substantially modifies a system or offers it under their own name can become a provider themselves.
Step by step, not all at once.
The AI Act applies in stages. Some rules, such as those on prohibited practices and AI literacy, already apply, others follow later, above all for high-risk uses. The deadlines were most recently adjusted by the so-called Digital Omnibus and may change further. Which ones apply right now is set out in the current version of the regulation.
Four topics for deployers.
Art. 4
AI literacy
Employees who work with AI should be able to develop the competence they need. Training and clear usage rules are an obvious place to start.
Art. 5
Prohibited practices
The regulation generally prohibits certain uses of AI, such as some forms of emotion recognition in the workplace. It is worth settling this in your own usage rules.
Art. 50
Transparency
In certain cases it should be recognisable that content was created with AI, for example deceptively real images, audio or video.
Art. 26
High-risk use
Where AI is used for decisions about people, for example in HR, further requirements may apply, among them human oversight and documentation.
Where a platform can help.
Whether a use of AI meets the requirements depends above all on what AI is used for and how. Software alone makes no one compliant. It can, however, help put your own rules into practice and make traceable what happened.
A log of every access
Who used which document with which model, and when, is in the audit log. That helps make processes traceable.
SecurityApproval by a person
In workflows a person decides at every point you define before anything moves on.
WorkflowsRoles and permissions
You decide who uses which assistants and data. That way usage rules can be reflected technically as well.
Sources for every answer
Answers name document and page. So employees can check a result before they use it.
Research
This page is meant as general orientation, is not complete and is not legal advice. Only the text of the AI Act in its current version is binding.
Questions about the EU AI Act.
Anything left open? In a free initial call the Ziya team answers your questions directly.
See Ziya One with your own data.
Live on your use cases, from the question to the finished file. 30 minutes, free and without obligation.