Banking & Fintech /

AI Act 2026: New responsibilities for high-risk AI- time to prepare

From August 2, 2026, further provisions of the EU Artificial Intelligence Act (AI Act) will come into force. These include, among others, obligations for so-called high-risk systems and the launch of regulatory sandboxes in each EU Member State.

From August 2, 2026, further provisions of the EU Artificial Intelligence Act (AI Act) will come into force. These include, among others, obligations for so-called high-risk systems and the launch of regulatory sandboxes in each EU Member State.

What is already in force?

Key provisions of the Artificial Intelligence Act of 2025, including:

  • general rules on the use of AI in the EU,
  • prohibitions on practices deemed unacceptable (unacceptable risk AI),
  • provisions on general-purpose AI models (GPAI).

These regulations are already affecting how companies can develop and offer AI-based tools.

From August 2026- high-risk systems and new obligations

The new stage of AI Act implementation will focus on so-called stand-alone high-risk AI systems, i.e., tools used in sensitive areas such as:

  • biometric recognition,
  • access to work, education, or public services,
  • assessment of candidates, loans, benefits.

The new stage of AI Act implementation will focus on so-called stand-alone high-risk AI systems, i.e., tools used in sensitive areas such as: biometric recognition, access to work, education, or public services, assessment of candidates, loans, benefits.

Who does the legislation apply to?

AI providers

Companies that bring high-risk AI systems to market will have to meet specific technical requirements, including:

  • data risk and quality management,
  • compliance documentation,
  • human oversight and accuracy testing mechanisms.

AI users (deployers)

Entities that use such systems (e.g., employers, institutions, recruitment companies) will be required to:

  • implementation of the supplier’s recommendations,
  • training of personnel,
  • ensuring adequate supervision and control of the system.

New responsibilities regarding transparency and testing of AI

The following will also come into force in August 2026:

  • Transparency requirements– e.g., informing users about interactions with AI,
  • The obligation for member states to create at least one regulatory sandbox– a testing environment for AI, with the participation of supervisory authorities.

Do you use AI in your company? Check whether you are subject to the provisions of the AI Act 2026.

Don’t wait until the last minute. It is worth identifying AI systems that may be subject to the new regulations now and implementing initial compliance and documentation measures.Contact us and subscribe to our newsletter to stay up to date with the latest regulations.

AI Act 2026

Stay up to date and subscribe to our channel!

Author team leader D&P Legal Natalia Chudzicka
Contact our expert
Write an inquiry: [email protected]
check full info of team member: Natalia Chudzicka-Szwajka

Contact us

Flaga Polski.POZNANPOLAND
pl. W. Andersa 3
61-894 Poznań
+48 61 853 56 48[email protected]
Flaga Polski.WARSAWPOLAND
Rondo ONZ 1
00-124 Warsaw
+48 22 300 16 74[email protected]
Flaga Polski.KRAKOWPOLAND
Opolska 110
31-355 Kraków
+48 61 853 56 48[email protected]
Flaga Polski.ZIELONA GÓRAPOLAND
Jana Sobieskiego 2/3
65-071 Zielona Góra
+48 61 853 56 48[email protected]
Flaga Włoch.MILANITALY
Via F. Sforza 15
20122 Milan
+48 61 853 56 48[email protected]