European AI Act: What Companies Must Anticipate from 2025

A Historic Turning Point for Artificial Intelligence in Europe

Adopted in 2024, Regulation (EU) 2024/1689, better known as the European AI Act, constitutes the world's first comprehensive legislation governing artificial intelligence. This text sets the rules of the game so that AI technologies are safe, transparent and respectful of fundamental rights.

Its objective is twofold: to protect citizens while encouraging responsible innovation. The European Union thus wishes to establish a climate of trust, essential for the massive adoption of AI in companies.

Source: Regulation (EU) 2024/1689 – Official Journal of the European Union, June 13, 2024.

A Risk-Based Approach

The heart of the system is based on a classification by risk levels:

• ▸Unacceptable risk: certain AI practices are prohibited (behavioral manipulation, social scoring, intrusive biometric recognition).
• ▸High risk: concerns systems used in sensitive areas (health, employment, education, security, personnel management) with strict obligations.
• ▸Limited risk: transparency obligation, for example informing that a chatbot is powered by AI.
• ▸Minimal risk: the majority of common uses, without particular constraints.

This approach allows protecting without blocking by adapting regulation to the criticality of each use.

(classification from ActiveMind Legal guide)

Who is Affected by the European AI Act?

The text applies to any public or private organization that:

• ▸Develops, places on the market, integrates or uses an AI system within the European Union;
• ▸Or offers AI products and services accessible to European users, even from abroad.

Any company that deploys or uses an AI system is therefore affected: from solution provider to simple user (HR, IT, business departments, data and innovation departments).

(source: Fasken, 2024)

Key Obligations to Anticipate

1. Risk Management

Companies must implement an AI risk management system covering the entire system lifecycle: design, testing, deployment, maintenance, withdrawal.

2. Documentation and Traceability

Each model must be documented: training data, conditions of use, levels of autonomy, human supervision processes. Objective: ensure the explicability of AI decisions.

3. Transparency and Human Supervision

Users must know they are interacting with an AI. For high-risk uses, human intervention remains mandatory to validate or correct automated decisions.

4. Conformity Assessment

Before any market placement, high-risk systems must undergo a conformity audit. European bodies will verify technical and documentary compliance.

(see Deloitte analysis, 2024)

5. Governance and Training

Companies must prove they master their AI uses: internal policy, designation of responsible persons, training plan and continuous updating.

Deadlines to Know

• ▸August 2024: entry into force of the regulation.
• ▸February 2025: application of prohibitions for unacceptable risk uses.
• ▸August 2025: specific obligations for general-purpose AI models (GPT, Gemini, Claude, etc.).
• ▸August 2027: full application for high-risk systems.

The first obligations apply from 2025: companies must prepare now.

(source: ArtificialIntelligenceAct.eu, 2024)

Why General Management and HR are on the Front Line

The AI Act is not just technical. It involves governance, ethics, cybersecurity and talent management. Compliance therefore requires direct involvement of HR departments and general management.

HR and Managerial Issues

• ▸Review AI uses in recruitment, evaluation and training.
• ▸Guarantee transparency and non-discrimination.
• ▸Train employees in supervision and AI governance.

Strategic Issues

• ▸Anticipate audits and documentation.
• ▸Identify compliant suppliers.
• ▸Adapt internal policies (security, GDPR, data governance).

Getting Compliant: Where to Start?

1. Map AI uses in your organization.
2. Identify the risk level of each system.
3. Document: datasets, supervision, incidents.
4. Train teams in responsible AI and compliance.
5. Update your internal policies and processes.

Training: Central Lever of Compliance

No compliance strategy works without human skills. Employees must understand the functioning, limits and risks of the AI tools they use.

Ascent Formation supports public and private organizations in their skills development on AI, cybersecurity and compliance. Our programs combine real cases, applied pedagogy and regulatory alignment.

The Ascent Approach: Regulatory Expertise and Applied Pedagogy

With references such as EDF, Thales, Orange, Total, Crédit Agricole or the Ministry of Armed Forces, Ascent Formation works on high strategic value projects. Our approach is based on three pillars:

• ▸Strategic understanding: deciphering the legal framework and its concrete implications.
• ▸Operationalization: translating AI Act requirements into internal action plans.
• ▸Training and support: training decision-makers, managers and employees in AI governance and compliance.

Our Recommended Training on AI and Compliance

Supporting Company Transformation in the AI Era

Equipping executives and managers to understand the strategic, technological and human impacts of AI on their organization.

Audience: executives, innovation managers, HR, IT.

Objective: build an AI adoption strategy compliant with the European framework.

Supporting AI Ambassadors in Their Mission

Giving AI referents the skills necessary to promote ethical, controlled and compliant AI.

Audience: managers, project managers, AI ambassadors.

Objective: manage AI transformation within teams, in connection with AI Act compliance.

Managerial Support and Change Management

Integrating the human dimension of AI compliance and supporting associated organizational transformations.

Audience: executives, HR managers, team managers.

Objective: guarantee buy-in and skills development in a regulated context.

Transforming Compliance into Opportunity

The AI Act is an opportunity to strengthen trust, transparency and responsibility in the use of artificial intelligence. Companies that prepare today will have a clear competitive advantage tomorrow.

(analysis KPMG, 2024)

Some European companies have expressed reservations about the application schedule, asking for more time to comply (Le Monde, July 4, 2025). However, deadlines remain fixed, hence the importance of anticipating now.

Contact Ascent Formation

Anticipate AI Act compliance today.

Our experts support companies in understanding, training and implementing the European regulatory framework on AI.

Contact us to build your customized training plan.

---