
How can you prepare your company for the new AI obligations through 2028?
European Artificial Intelligence Act (AI Act)
The European Artificial Intelligence Act (AI Act), established by Regulation (EU) 2024/1689, which lays down harmonised rules on artificial intelligence, is now generally applicable, although some of its obligations have been in force since 2025 and others will be phased in progressively until 2028. This AI Act implementation timeline requires companies to understand which requirements apply to them and when they must comply.
The regulation does not apply solely to those who develop or market AI systems: it also covers organisations that use them in their professional activities. Below, we review the AI Act obligations for companies that are already enforceable, the upcoming milestones and the measures that should be planned to ensure compliance.
Which AI Act obligations should your company have already reviewed?
Since February 2025, companies have been required to take measures to ensure that employees using artificial intelligence tools understand how they work, their limitations and the associated risks. This artificial intelligence training must be tailored to each professional's roles and responsibilities, as not all uses of AI pose the same risks. In addition, certain practices that may affect individuals' rights have been prohibited since that date, such as certain systems designed to manipulate behaviour or, subject to the exceptions provided for by law, those used to infer emotions in the workplace.
Furthermore, specific obligations for providers of general-purpose AI models have been enforceable since August 2025, while certain AI Act transparency requirements have applied since August 2026. These may include, among other measures, informing individuals when they are interacting with AI systems or identifying artificially generated content. It is therefore essential for companies to review the tools they use and determine which obligations fall on their providers and which they must fulfil directly as users of these systems.
What are the upcoming key dates for the AI Act?
2 December 2026 – New prohibitions and labelling of AI-generated content
New prohibitions relating to certain unlawful applications come into effect, and a transitional period for the technical marking of AI-generated content by certain pre-existing generative systems comes to an end. Companies should consult their providers about how they identify synthetic content and review the tools incorporated into their communication or customer service processes.
2 December 2027 – Obligations for high-risk AI systems
Requirements for certain high-risk artificial intelligence systems in sensitive areas, such as recruitment and employee assessment or the evaluation of individuals' creditworthiness, come into effect. Depending on the system and the company's role, human oversight, record-keeping, information provision and other control measures may be required. Not every application used in these processes is automatically classified as high-risk: its specific intended purpose is what matters.
2 August 2028 – New requirements for AI integrated into regulated products
The high-risk regime is extended to certain AI systems integrated into regulated products, such as certain medical devices or toys. Affected manufacturers and providers will need to incorporate these requirements into their conformity assessment processes.
How can you prepare your company to comply with the Artificial Intelligence Act?
The first step is to create an inventory of the artificial intelligence tools in use, including services contracted by each department. It is advisable to record what they are used for, who oversees them, and whether they process personal data or confidential information or are involved in significant decisions. This comprehensive overview helps identify uses that do not always come under the supervision of technical teams.
Next, it is important to determine whether the company acts as a provider or a deployer in each case, review the terms of contracts with third parties, and establish clear guidelines on what information can be entered into the tools. Where personal data is processed, compliance with the AI Act and the GDPR must be coordinated: one regulation does not replace the other.
Finally, identifying systems that may be classified as high-risk, training teams and assigning internal responsibilities helps prioritise actions. Preparing in advance avoids having to make rushed adjustments to human resources, procurement, technology or customer service processes.
AI Act Implementation Timeline from 2025 to 2028
| Date | Obligation or Regulatory Milestone |
|---|---|
| 2 February 2025 | AI training and implementation of certain prohibitions |
| 2 August 2025 | Obligations for providers of general-purpose AI models |
| 2 August 2026 | General application and transparency requirements |
| 2 December 2026 | New prohibitions and the end of certain transitional periods |
| 2 December 2027 | Requirements for high-risk AI systems in sensitive areas |
| 2 August 2028 | Requirements for high-risk AI systems integrated into regulated products |
Why prepare for AI Act compliance in advance?
The AI Act implementation timeline presents an opportunity to strengthen trust in the use of technology and move forward with a well-informed approach.
At Baker Tilly, our Data Protection and Corporate Compliance teams help your company identify the obligations that apply to it, assess the risks associated with its AI tools and develop a realistic AI Act compliance roadmap aligned with its processes and business objectives.
Because innovating safely starts with being prepared.