The transparency obligations from the European AI Act have officially been in effect since August 2. What should you as an organization take into account?
The European Union is taking the next step in the regulation of artificial intelligence (AI Act). Since August 2, new transparency obligations have been in effect that require companies to make certain AI-generated content recognizable as such. The rules (Article 50) apply to companies, governments, and the self-employed who deploy AI. Fines can reach up to 15 million euros or three percent of global turnover. What do the new measures mean for you as an organization?
European AI Act
The European AI Act is no longer the distant future, but an imminent reality. Although the EU AI Act already entered into force on August 1, 2024, the obligations are trickling in gradually. The full rollout is expected to take place before August 2, 2028, according to the implementation timeline.
That same timeline also shows a new milestone. Since August 2, 2026, most rules of the AI Act have come into effect. Not only do the transparency rules from Article 50 apply, but enforcement of the AI law has also started at national and European levels. This means that companies can now be officially monitored and fined if they violate the rules.
Article 50
Article 50 of the European AI Act focuses on transparency. The measures are intended to ensure that users know when they are dealing with AI and that artificially generated content is labeled as such.
The measures target everyone who uses AI professionally: companies, governments, the self-employed, and professional influencers fall under the rules, while purely private use generally remains outside their scope.
Anyone using a chatbot must inform the person concerned from the first contact that they are talking to AI, and realistic AI content in image, audio, or video that could pass for real must be visibly or audibly identified as artificial.
For AI text, a mandatory label applies to publications concerning the public interest, except when there is genuine human editorial control and final responsibility. For AI systems that create synthetic audio, image, video, or text, there is also an obligation to mark this output as artificial in a machine-readable format.
From building AI to governing AI
The new transparency rules force companies to scrutinize their AI usage. Where the focus was initially on rapid AI adoption, it is now shifting toward a thorough analysis of AI use within companies.
“With the AI Act, the focus shifts from building AI to governing AI,” says Luis Blando, Chief Product & Technology Officer at OutSystems. “Now that AI agents are increasingly supporting business-critical processes, insight into their operation, decision-making, and the moments when human oversight is needed becomes essential.”
With the AI Act, the focus shifts from building AI to governing AI.
Luis Blando, Chief Product & Technology Officer bij OutSystems
Blando points out that nearly two-thirds of organizations do not yet have a centralized approach to AI governance. This is partly due to the use of outdated IT architectures, which offer insufficient visibility and control.
“Governance cannot, therefore, be added as an extra layer afterward. It must be part of the architecture from the start, so that organizations can centrally manage, monitor, and orchestrate AI systems, even when they are deployed at scale,” says Blando.
Trust in the digital foundation
In addition, AI is increasingly becoming part of the digital foundation that companies rely on daily, according to Bilal Ahmed, Chief AI & Data Officer at team.blue. “Just as the General Data Protection Regulation (GDPR) created a common framework for handling personal data responsibly, the AI Act lays the foundation for trustworthy AI in Europe.”
“For SMEs, it’s not just about compliance with legislation. It’s about being able to trust that the AI they use is transparent, responsible, and fit for its intended purpose. The AI Act should not slow down the adoption of AI, but rather encourage companies to use AI more consciously.”
Is Belgium ready?
Although the first measures are in effect, uncertainty remains regarding Belgium’s readiness. According to the Central Economic Council (CCE), Belgium is currently not yet ready to implement the AI Act correctly. The CCE writes that no national supervisor has been appointed yet. Furthermore, there is still too much lack of clarity regarding how federal states will distribute competencies. According to the CCE, Belgian governments must first arrive at a clear regulatory framework.
Regarding the use of AI among Belgian companies, the report is remarkably positive. About a third of Belgian employees and companies (34.5%) use some form of AI technology today. That puts Belgium well above the European average.