AI Act Compliance

Mapping your AI systems, risk assessment, AI governance, and alignment with the GDPR: we help you structure your compliance with the European AI Regulation to ensure the long-term security of your AI uses, well before the high-risk deadline of December 2027.

100% Satisfaction
13 Reference systems
+250 Customers
Square with text bubble Complies with requirements

More than 250 companies have already placed their trust in us

Logo aniah
jamespot logo
Logo airon telematica
Logo seqino
SBS Interactive logo
auqfood logo

Understanding the AI Act
what it means for your business

The AI Act (EU Regulation 2024/1689) is the world’s first legal framework governing artificial intelligence. It took effect on August 1, 2024, and will be phased in through 2028. It applies to any organization that develops, markets, imports, distributes, or uses an AI system in the European Union, including non-EU companies whose AI systems affect individuals in Europe.

The regulation is based on a risk-based approach: each AI system (AI system) is classified into one of four categories (unacceptable, high risk, limited risk, minimal risk). The AI Omnibus Regulation (EU 2026/1744), in effect since July 27, 2026, has deferred the obligations for high-risk AIS, but the prohibitions, transparency requirements, and AI oversight provisions are already in effect. Key initiatives (mapping, governance, documentation) generally take 4 to 7 months to complete.

Unlike the GDPR, the AI Act does not require an “AI DPO,” but it does require dedicated governance: a mapping of your AI systems, a designated lead, and AI control measures for your teams. When done right, compliance becomes an asset in your RFP processes and safeguards your AI projects.

Risk-Based Approach (4 Levels)
Each AI system is capable of applying the correct treatment regimen
Measures to train teams that use AI
AI Governance (Article 4)
Coordination with the GDPR
AIPD is required for most high-risk information systems that process personal data
AI Act
EU Regulation 2024/1689 · Amended by the AI Omnibus Act (EU 2026/1744)
35 M€ Or 7% of global revenue — maximum penalty (prohibited practices)
4 levels of risk to classify each AI system
Dec. 2, 2027 Application of Requirements for High-Risk SIA (Annex III)
27 countries Uniform application throughout the EU
Any company that uses a generative AI tool, an algorithmic scoring system, or automated sorting is affected, even if it merely deploys an AI tool provided by a third party.

Are you a supplier
or an AI implementer ?

The AI Act distinguishes between several roles (supplier, deployer, importer, distributor, and agent). For the vast majority of small and medium-sized enterprises (SMEs) and mid-sized companies, the key issues center on two of these roles: this classification determines the nature and scope of your obligations.

Case No. 1

AI System Provider

You develop (or have developed) an information system and market it under your name or brand · penalties of up to €15 million or 3% of global revenue (high-risk information system)

Your Main Responsibilities

  • Lifecycle Risk Management (Art. 9)
  • Data Governance and Quality (Art. 10)
  • Complete technical documentation (Art. 11)
  • Transparency and Instructions for Use (Art. 13)
  • Quality Management System (Art. 17)
  • Conformity Assessment & CE Marking
  • Registration in the EU database
  • Post-Market Surveillance
Case No. 2

AI System Deployer

You use an AI system in your business: generative AI, resume screening, scoring, chatbots · this is the case for the majority of small and medium-sized businesses and mid-sized companies · penalties of up to €15 million or 3% of global revenue

YOUR MAIN OBLIGATIONS

  • Use in accordance with the supplier's instructions
  • Effective human oversight by trained personnel
  • Operational Monitoring and Log Retention (Art. 26)
  • Information for Employees and Affected Individuals
  • Transparency Toward Users (Chatbots, User-Generated Content)
  • AIPD under the GDPR if personal data is involved
  • Fundamental Rights Impact Assessment (certain network operators, Art. 27)
  • AI Control Measures for Your Teams (Art. 4)

Just a user, so not affected? That’s not true. As a deployer, you have real operational responsibilities, especially for high-risk AI systems. And if you substantially modify a third-party AI tool or integrate it into your product under your own brand, you may be classified as a provider. To determine your role, schedule a 30-minute AI Act assessment with one of our consultants.

The risk levels defined by the AI Act

The AI Act classifies each AI system based on the risk it poses to health, safety, and fundamental rights. Each level triggers a specific set of obligations.
In addition, there are two cross-cutting frameworks: general-purpose AI models and AI governance.

Unacceptable risk: prohibited

Practices contrary to EU values, prohibited as of February 2, 2025: social scoring, cognitive manipulation, exploitation of vulnerabilities, and emotion recognition in the workplace. New prohibitions take effect on December 2, 2026. Fines of up to 35 M€ or 7% of global revenue.

High risk (Annexes I and III)

Resume screening, employee evaluation, credit scoring, education, critical infrastructure: risk management, data governance, technical documentation, human oversight, CE marking. Effective December 2, 2027 (Annex III), and August 2, 2028, for AI integrated into regulated products (Annex I).

Limited risk: transparency

Chatbots, virtual assistants, and content generated or modified by AI: Users must be aware that they are interacting with AI. The requirements of Article 50 take effect on August 2, 2026, and content generated by AI must be labeled by December 2, 2026, for systems already on the market.

Minimal risk: unrestricted

Spam filters, productivity tools, simple recommendation engines: no binding requirements under the AI Act; voluntary codes of conduct are encouraged. This covers the vast majority of business uses, provided you have classified them in your mapping.

General-Purpose AI (GPAI) Models

As of August 2, 2025, providers of GPT-type models—such as Claude, Mistral, or Gemini—have specific obligations (transparency regarding training, copyright, and systemic risks). Asa user of these tools, you remain responsible for using them appropriately in your business.

AI Governance (Article 4)

Vendors and implementers must take steps to develop their teams’ AI proficiency. Effective as of February 2, 2025, this requirement has been relaxed to a “best-efforts” obligation under the AI Omnibus Act: it still must be implemented and documented.

Good to know. A single company often uses AI systems with varying risk levels without even realizing it. An incomplete assessment leaves you vulnerable to audits, formal notices, and financial and reputational penalties. A 30-minute assessment with one of our consultants is all it takes to evaluate your AI systems and prioritize improvement projects. Want to discuss this?

AI Act: A High-Stakes Issue
—Still Underestimated by Companies

Most companies believe they are not affected by the AI Act because they do not develop AI. In reality, virtually all small and medium-sized businesses are using AI, often without even realizing it.

Shadow AI: Your Biggest Blind Spot

ChatGPT, Copilot, Gemini, Notion AI… Your teams use generative AI on a daily basis, often without any usage policies or audit trails. Your company remains legally liable for this use.

No mapping of AI systems

Resume screening, customer scoring, chatbots, predictive models: Most companies don’t know how many information systems they actually use, or the level of risk associated with them. Mapping these information systems is the first step toward achieving compliance.

AI providers that have never been evaluated

Your SaaS vendors are integrating AI into their products without always documenting their compliance. As a deployer, you need to obtain clear documentation and contractual assurances regarding the systems you use.

An informal approach to AI (Article 4)

As of February 2025, you must take steps to improve your employees’ proficiency with AI: raising awareness, establishing a code of conduct, and providing role-specific training. In most companies, nothing has yet been formalized or documented.

The Confusion Between the AI Act and the GDPR

Many people think that complying with the GDPR is enough. That is not true:the AI Act adds its own requirements (mapping of AI systems, technical documentation, human oversight, logging, transparency) that supplement those of the GDPR.

A schedule that was imposed rather than planned

Transparency requirements effective as of August 2026, new prohibitions effective as of December 2, 2026, and high-risk status effective as of December 2, 2027: waiting until the final deadline risks slowing down or halting AI projects that are already underway.

FeelAgile turns these obstacles into manageable steps.

Our agile approach turns these obstacles into manageable steps. A dedicated expert guides you through each milestone, simply and effectively.

Talk to an expert →

Why Choose Us for Your Compliance
AI Act ?

FeelAgile brings together legal professionals, DPOs, and cybersecurity experts trained in the requirements of the AI Act. Our approach draws on our experience with the GDPR, ISO 27001, ISO 42001, and HDS, adapted to the AI framework: each requirement is supported by tools, documented, and auditable.
Expert guidance

ISO 42001 as a Governance Framework

The ISO 42001 AI management system provides the operational framework for meeting the governance, risk management, and documentation requirements of the AI Act.

A Combined Legal, IT, and Data Team

Data privacy lawyers, cybersecurity experts, and data professionals work together to ensure legal, technical, and operational compliance.

A Pragmatic Approach for Small and Medium-Sized Enterprises (SMEs) and Mid-Sized Companies

No over-engineering: Our deliverables are tailored to your actual AI maturity level, the number of AI systems in production, and your role as a deployer or provider.

Optimized lead times

From assessment to operational compliance in less than 6 months, starting by ensuring compliance with existing requirements.

An agile approach aligned with the GDPR

We build on your GDPR register, your AIPDs, and your security policies to avoid duplicate documentation: AI governance is an addition, not an overlay.

A Comprehensive Approach to Compliance

AI Act, GDPR, ISO 42001, ISO 27001, NIS 2, HDS: We build coherent governance systems by consolidating controls, policies, and evidence.

Make ISO 42001 the foundation of your AI Act compliance
The ISO/IEC 42001 standard provides a framework for an Artificial Intelligence Management System (AIMS): AI policy, risk and impact analysis, system lifecycle, and supplier management. Certifiable by a third-party organization, it is the preferred standard for demonstrating robust and auditable AI Act compliance.

Our ISO 42001 Services →

They trust us

More than 250 clients have already used FeelAgile

★★★★★

"Thanks to Feel Agile, we managed to achieve ISO 27001 certification without any non-conformities, which is a rare feat."

Profile photo Julien Caasagnabere
Val Solutions

Julien Cassagnabère -RSSI

★★★★★

"We received excellent support. The project manager thoroughly reviewed our quality system, which made the entire project run smoothly."

Male image
Airon Telematica

Stefano FIORENTINI - CTO

★★★★★

"Feel Agile has a deep understanding of the process, a project plan with an efficient timeline, and existing documentation to save time."

Male image
Aniah

Mickaël KLAUS

Expert on these regulations

DORA logo
GDPR logo
NIS2 logo
But also ISO 42001, the Cyber Resilience Act, and the Data Act
Read more

Learn more about AI AI Act

Guides, webinars, and support to help you understand your obligations, plan ahead, and organize your internal process.

Complete Guide

The ISO 42001 Guide: Governing AI in the Workplace

Understand the AI management system standard, how it relates to the AI Act, and the key steps toward responsible, ethical, and controlled AI.

View the guide →

📖 Reference Guide

Webinar

AI in the Workplace: How to Secure Your AI Usage from Start to Finish

The AI Act, ISO 42001, and usage governance: the keys to regulating generative AI and AI systems in your organization.

Watch the webinar →

🎥 Replay available

Support

ISO 42001 Certification: Our Support

Risk and Impact Analysis, SMIA, Mock Audit: Our Approach to Structuring the Governance of Your AI Systems and Working Toward Certification.

Explore the offer →

🧠 ISO/IEC 42001

FAQ

Questions Frequently Asked Questions

Everything You Need to Know About AI Act Compliance

Which companies are affected by the AI Act?

Any organization that develops, places on the market, imports, distributes, or uses an AI system in the European Union, including companies located outside the EU whose AI systems affect individuals in Europe. Virtually all small and medium-sized enterprises (SMEs) and mid-sized companies are affected as implementers: generative AI tools, resume screening, scoring, and chatbots. The requirement to ensure control over AI (Article 4) has applied since February 2, 2025, to any organization that uses an AI tool in a professional context.

What is the timeline for implementing the AI Act?

The regulation took effect on August 1, 2024, and is being implemented in phases: February 2, 2025 (prohibited practices and AI control), August 2, 2025 (general-purpose AI models), August 2, 2026 (transparency requirements under Article 50, governance, and penalties), December 2, 2026 (new prohibitions and labeling of content generated by systems already on the market), December 2, 2027 (high-risk AI systems listed in Annex III), and August 2, 2028 (AI integrated into regulated products, Annex I). These last two dates result from the Omnibus AI Regulation (EU 2026/1744), which has been in effect since July 27, 2026.

What penalties apply in the event of noncompliance?

Fines are scaled according to the severity of the violation: up to 35 M€ or 7% of global revenue for prohibited practices (Article 5); up to 15 M€ or 3% for most other obligations, particularly those of high-risk SIA; and up to 7.5 M€ or 1% in cases where inaccurate information is provided to the authorities. For SMEs and startups, the cap is set at the lower of the absolute threshold or the percentage of revenue. Beyond the fine itself, the reputational and commercial risks are often the primary concern.

Am I an “AI provider” or an “AI implementer”?

You are a provider if you develop an AI system (or have it developed) and market it under your own name or brand. You are a deployer if you use an AI system provided by a third party in your professional activities: this applies to the vast majority of small and medium-sized businesses and mid-market companies that use ChatGPT, Copilot, or resume-screening and scoring tools integrated into their HRIS or CRM systems. Please note: if you substantially modify a third-party AI system or integrate it into your own product under your brand, you may be classified as a provider, which entails stricter obligations.

Does my company need to appoint an “AI DPO” or an AI Officer?

Unlike the GDPR, the AI Act does not formally require the appointment of a dedicated role. However, compliance requires structured governance: a point person equipped with the necessary resources, a solid understanding of the regulation, and a knack for technical matters. In practice, the existing DPO is often well-positioned to take on this role, with some additional training. FeelAgile can support your DPO or lead the process alongside you.

How long does it take to become compliant?

For an organization that uses a few AI systems—few or none of which are high-risk—you can generally expect it to take 4 to 7 months to achieve operational compliance. Our approach begins with a 3- to 4-week assessment and mapping phase, which first ensures compliance with existing requirements (prohibited practices, transparency, AI governance) and then structures the rest of the process into short sprints.

How do the AI Act and the GDPR relate to each other?

Both regulations apply concurrently. Any AI system that processes personal data remains fully subject to the GDPR: record-keeping, legal basis, retention period, data subject rights, and security. For high-risk AI systems that process personal data, a Data Protection Impact Assessment (Article 35 of the GDPR) is generally required: it complements the AI Act risk analysis. Leveraging your existing Data Protection Impact Assessments is the most effective way to reduce the administrative burden. Learn more about our GDPR support services.

AI Act, ISO 42001, ISO 27001, NIS 2: How Do These Standards Relate to One Another?

They are complementary.The AI Act is the European legal framework for AI.ISO 42001 is the international standard for AI management systems, the preferred framework for demonstrating operational compliance.ISO 27001 provides the foundation for information security, which can be leveraged to meet the robustness and cybersecurity requirements of AI systems. NIS 2 mandates cyber risk management and the reporting of incidents to the relevant entities. FeelAgile builds an integrated governance system that covers these standards without redundant documentation. Learn more about our ISO 42001 support services.

Our articles

Want to learn more from our articles on AI and compliance?

Our experts will get back to you within 24 hours.

Are you using generative AI tools without a usage policy, or integrating AI into your products? Tell us about your situation, and a FeelAgile consultant will call you back.

More than 250 companies trust us
jamespot logo
auqfood logo
SBS Interactive logo
Logo seqino
Logo aniah
Logo airon telematica