The effects of the AI Act on companies – "Keep a cool head"

The EU's AI Act is the world's first comprehensive set of AI regulations. Its essence is simple. The aim is to steer the development and use of AI in a way that is safe, transparent, and respectful of human rights.

Therefore, AI applications are evaluated based on the risk they pose to users. There are four risk levels, each with varying regulatory requirements. Often, the risk is minimal, but it still needs to be assessed.

The Digital Omnibus published by the European Commission in November 2025, on the other hand, proposes targeted amendments to the AI Regulation. The aim is to simplify the rules and reduce companies' compliance obligations and associated costs.

"If you have done things smartly so far, I would argue that the effects of regulation are not as great as many people have thought in many use cases," says Juhana Juppo, Digia's Chief Technology Officer.

Juppo says that he encounters even wild notions about what the regulation prohibits.

"Many people seem to believe that the AI Act will make the use of AI impossible or very difficult. This is not quite true," Juppo says.

Only one of the risk categories prohibits the use of AI.

"Solutions that violate human rights and are defined as an unacceptable risk are those that would be against the law or at least against ethics and the general sense of justice. I do not believe that this ban at the EU level will have much impact on how companies in Finland use or plan to use artificial intelligence," Juppo points out.

In other categories, including the high-risk category, AI use is allowed. You just need to know what you're doing and follow the instructions in the category.

"Of course, there are a lot of requirements for an organization in the high-risk category, but if you have acted smartly and with high quality, i.e., assessed the risks, documented your actions, told users about the use of AI, followed ethical principles, and have the processes in order, the change will not be that big," Juppo says, and continues:

"If, on the other hand, these things are not in order, you should start working right away."

According to Juppo, it is possible and worthwhile to proceed boldly with AI while accounting for risk assessments. Common sense goes a long way.

"If there's a bridge in front of you and you doubt whether it will last, you can choose whether to check first or press forward with your eyes closed and see what happens. So that's how it goes in private life as well: assess the risks and be prepared to endure the consequences of your choice," Juppo adds.

The full application of the EU AI Act started gradually at the beginning of 2025 and will continue until 2027.

• The AI Act entered into force on 1 August 2024.

• The regulation will be implemented in stages under different transitional periods 

• The rules on prohibited use cases entered into force on 2 February 2025. 

• In November 2025, the Digital Omnibus was published, which proposed targeted changes to the EU AI Regulation and its timelines

• In June 2026, the EU announced upcoming changes to the AI Act and new transition periods

• Mandatory AI content tagging will enter into force in December 2026

• According to the new schedule, the ban on high-risk AI systems will not enter into force until December 2027 (instead of August 2026)

• The Council of Europe and the European Parliament will have to approve the upcoming amendments to the legislation before finalizing and adopting it (as of June 2026)