The recent entry into force of the transparency provisions of the European AI Regulation – known as the AI Act – represents a turning point for large developers of AI models such as OpenAI, Anthropic, Google, Meta or Mistral. But for SMEs and technological freelancers, this regulation also marks challenges and opportunities that should be understood and taken advantage of rigorously.
Why does this matter for SMEs and tech freelancers?
Firstly, although the regulations are designed especially for providers of general purpose models - such as OpenAI with ChatGPT or Google with Gemini - what they do has a direct impact on those of us who use these models to offer services. If these large players change licenses, documentation or transparency, SMEs that integrate AI in customer service, marketing or content production may be affected.
On the other hand, these transparency obligations foster a safer and more predictable environment: freelancers who offer AI-based services will be able to communicate more clearly to their clients when they are interacting with an AI or when texts, images or audio are artificially generated. In short, trust and differentiation for smaller companies.
- Clarity in front of the customer: know when AI is used in processes and products.
- Legal compatibility: integrative SMEs rely on more transparent and aligned models.
- Competitive advantage: Offering compliant AI-based services builds trust.
Context and timeline of the AI Act
The AI Act formally entered into force on August 1, 2024 (commission.europa.eu), promoting the first European regulatory framework for safe AI that respects fundamental rights. The regulation is applied in a phased manner, taking into account the different levels of risk and types of systems.
Specific obligations for general purpose models (GPAI) began to apply on August 2, 2025: from that date, providers must disclose key information about training data, respect copyright and provide technical documentation, supported by a code of good practices (interoperable-europe.ec.europa.eu).
August 2, 2025: Transparency obligations for GPT-like models begin.
August 2, 2026: Article 50 transparency rules are generally applied and active oversight by national authorities and the EU AI Office begins (digital-strategy.ec.europa.eu).
December 2, 2026: The grace period for labeling artificial content on models released before August 2026 ends (digital-strategy.ec.europa.eu).
Practical implications for suppliers and deployers
Article 50 of the AI Act imposes clear obligations on both providers – who develop the models – and deployers – who implement them into final services or products. Among the main requirements are:
- Inform the user when their interlocutor is an AI system.
- Label content generated or manipulated by AI as such, in a clear and mechanizable way.
- In the case of biometric technologies or emotional recognition, explain how they work and respect data protection regulations.
- Detect and label deepfakes or artificially generated texts, especially if there is no human editorial review.
Additionally, GPAI model providers must document their model in detail, including training data summary, copyright policies, systemic risk assessments, and report relevant incidents to the European Commission (openai.com). This allows SMEs that use these models to receive reliable information prepared to comply with regulations, without having to develop it from scratch.
Recommended steps for SMEs and self-employed workers
For SMEs and self-employed workers who use AI in their daily operations, these recommendations are key:
- Verify that your model provider is AI Act compliant: make sure they have documentation, clear labels, and transparency in line with regulatory dates.
- Communicate clearly to your customers: inform when AI is used, if the content is generated and how it has been verified.
- Design your own internal processes: Make sure you have human editorial review if you generate content for third parties.
- Stay up to date on regulatory developments: especially the December 2026 deadlines and the evolution of the GPAI code of good practice.
- Document your workflows – this will facilitate any future audits or legal requirements and improve customer confidence.
In an evolving regulatory environment, SMEs that proactively adopt these changes will gain professionalism, security and positioning against those who experience uncertainty or non-compliance.
In short, the AI Act not only affects giants such as OpenAI, Anthropic, Google, Meta or Mistral; It also redefines the standard of transparency in AI systems throughout the EU, becoming a strategic factor for small technology companies that are committed to growing with confidence.
Do you want CADT Software to accompany you in adapting your AI solutions to this new regulation? We can help you integrate transparent and responsible models under the AI Act.
