AI is no longer a future technology — it's part of everyday business operations.
Chatbots, document processing, HR automation, generative AI, and decision-support systems are appearing at more and more companies. At the same time, the first comprehensive European AI regulation has arrived: the EU AI Act.
The regulation fundamentally changes how artificial intelligence can be developed, deployed, and used in the European Union. The question is no longer whether your company uses AI, but whether it complies with the AI Act's requirements.
In this article, we summarize:
what the EU AI Act is,
which companies it affects,
which AI systems count as high-risk,
which deadlines to watch,
and how a local AI solution, such as Cognexa, can help with compliance.
What is the EU AI Act?
The EU AI Act (Regulation (EU) 2024/1689) is the European Union's first comprehensive AI regulation. The regulation entered into force on August 1, 2024, and is becoming fully applicable in stages.
The regulation aims to:
make AI systems safer,
increase transparency,
protect users,
while continuing to support innovation.
One of the most important features of the AI Act is that it takes a risk-based approach. What matters isn't the algorithm itself, but what it's used for.
The 4 risk categories of AI systems
1. Unacceptable risk – prohibited AI systems
The EU bans certain AI solutions outright.
These can include, for example:
systems built around manipulation,
psychological manipulation targeting children,
certain biometric surveillance systems,
emotion recognition in the workplace,
social scoring systems.
Using such systems can result in serious fines.
2. High-risk AI systems
This category includes, for example:
AI systems for HR and recruitment,
credit scoring systems,
AI for medical diagnostics,
educational assessment systems,
AI managing critical infrastructure.
For these, the following may be mandatory:
risk management,
documentation,
logging,
human oversight,
data quality assurance,
compliance audits.
3. Limited-risk AI
Examples include:
customer service chatbots,
generative AI systems,
AI-based content-generation tools.
The main requirement here is transparency:
users must be informed that they are interacting with AI,
AI-generated content must be labeled as such.
4. Minimal-risk AI
Most internal corporate AI tools fall into this category:
knowledge-base assistants,
document search tools,
internal automations,
analytics systems.
These carry minimal obligations.
Who does the EU AI Act affect?
Far more companies than you'd think at first.
The AI Act doesn't just affect AI developers — it affects every organization that:
uses AI for business purposes,
integrates an AI system,
uses an AI-based SaaS solution,
offers AI to its customers,
processes the data of EU users.
Typical affected areas:
HR,
customer service,
marketing automation,
financial scoring,
healthcare,
education,
compliance.
Key AI Act deadlines
February 2, 2025
These come into effect:
the rules on prohibited AI practices,
the AI literacy obligations.
August 2, 2025
The designated supervisory authorities begin operating.
August 2, 2026
The main body of the AI Act becomes fully applicable, in particular the rules governing high-risk systems.
What does a Hungarian company need to do?
1. Create an AI inventory
The first step is mapping out:
which AI tools the company uses,
which SaaS systems include AI features,
which processes are automated.
Many organizations use AI without having full visibility into it.
2. Risk classification
Every AI system needs to be classified as:
prohibited,
high,
limited,
or minimal risk.
3. Setting up AI governance
It's worth designating:
an AI officer,
a compliance team,
an internal AI policy,
supplier review processes.
4. Aligning GDPR and AI
The AI Act doesn't replace the GDPR.
If the AI processes personal data, you may still need:
a privacy notice,
a DPIA,
an appropriate legal basis,
access controls.
Why can local AI be advantageous in the age of the AI Act?
One of companies' biggest challenges is AI transparency and data security.
With cloud-based AI solutions, the following are often open questions:
where the data is stored,
which models the provider has access to,
how confidential documents are handled.
A local AI system, by contrast:
runs on your own infrastructure,
keeps data under your control,
is easier to audit,
fits more easily into compliance processes.
How can Cognexa help with AI Act compliance?
Cognexa is a local, document-based AI platform built on corporate knowledge bases and internal documents.
With it, you can:
make internal policies searchable,
link compliance documents together,
review AI processes more quickly,
support AI governance.
Cognexa's advantages:
Local operation
Sensitive corporate data never reaches public AI systems.
Document-based AI
The system is built on your own corporate knowledge.
Compliance support
Helps document and review AI Act and GDPR requirements.
Transparent operation
AI use that's easier to audit and control.
Summary
The EU AI Act isn't just a new piece of technology regulation. It sends companies a clear message:
Using AI is now also a compliance matter.
In the coming years, the companies that gain the advantage will be the ones that:
build transparent AI processes,
keep data under control,
establish deliberate AI governance,
and are able to operate secure, auditable AI systems.
Local AI solutions and platforms like Cognexa can play a key role in this.
Want to get your company ready for the AI Act era?
Cognexa's team can help with:
AI-based processing of internal documents,
building local AI systems,
supporting compliance and AI governance.