Careless corporate use of artificial intelligence can also violate trade secrets

By: Trademagazin Date: 2025. 12. 15. 10:57

🎧 Hallgasd a cikket: Nyelv: Sebesség: 1.05×

The Hungarian law on artificial intelligence (AI) came into force on December 1, which allows companies to verify whether they are legally using AI systems, the international law firm Baker McKenzie points out. The preparation period has already begun, and in case of improper use, the fine can reach 7 percent of sales revenue.

The European Union – as part of its digital strategy – regulates artificial intelligence (AI) in order to provide better conditions for its development and use and reduce its potential risks. The European Commission proposed regulating the area in 2021, and the Artificial Intelligence Regulation (AI Act) entered into force on August 1, 2024, after being approved by the EU Council and the European Parliament. The EU regulation classifies AI systems into four categories, from systems posing unacceptable risks to solutions with minimal risk. Stricter rules apply to solutions with greater individual user and social risk.

The regulation is also binding in Hungary – i.e. it does not need to be separately incorporated into the Hungarian legal system – and the Hungarian law on its implementation creates the institutional background for the application of the law. Most of its provisions have now entered into force on December 1. The law and its implementing decree designate the domestic authorities responsible for implementation. The AI notifying authority is the National Accreditation Authority, which designates and supervises organizations that assess the legal compliance of AI systems. The AI market surveillance authority is the Ministry of National Economy, which is responsible for business development, and which monitors compliance with the law, i.e. the legal use of the systems. This authority can impose sanctions, such as fines – this can be up to 15 million euros or 3 percent of sales revenue. A significant fine can be imposed for the use of prohibited systems that pose an unacceptable risk: this can reach 35 million euros or 7 percent of sales revenue. These include, for example, manipulation and subconscious techniques, as well as social scoring based on behavior, socioeconomic status or personal characteristics.

“According to the law, companies must ensure the proficiency of their employees. They must be informed about the possibilities, risks and damage that AI can cause. All this will require continuous training, as the field is developing rapidly: between the negotiation and adoption of the EU regulation, services based on large language models such as ChatGPT appeared, and the lessons learned from this were continuously incorporated into the law,” said Dr. Csaba Vári, head of Baker McKenzie’s IPTech practice.

In addition, it is worth making an inventory of what and what artificial intelligence is used by companies, and whether its use is necessary at all. Special attention must be paid to on the security issues of AI systems.”

According to the law firm’s experience, the preparedness of Hungarian companies varies. Large companies are primarily concerned with creating AI skills, not only because of the legal obligation in force since February. A significant challenge is the so-called shadow AI, when employees use artificial intelligence without approval, often sharing sensitive company data – this can cause significant damage, as it is not possible to know where this data ends up. Therefore, it is important to raise internal awareness so that business secrets or personal data are not leaked through the use of AI. It is also clear that cybercriminals also work with artificial intelligence, and recognizing this poses a serious challenge for employees.

The recently adopted law also provides for the creation of a test environment – this is also being developed by the ministry responsible for market supervision. Here, companies developing artificial intelligence can test in a regulated environment whether their products comply with the AI regulation. This part of the provision will enter into force on 2 August 2026.

In order to better prepare companies and simplify the rules, especially for SMEs and medium-sized companies, and reduce the administrative burden, the European Commission published a so-called digital omnibus package in November, covering several pieces of legislation. This would, among other things, postpone the entry into force of the provisions on high-risk systems until the end of 2027 at the latest. However, the Commission’s proposal still needs to be adopted by the European Parliament and the Council, so companies should keep an eye on any changes to the provisions that apply to them.