AI Act Compliance for Software Vendors | Rakam AI
Back to Blog
Regulation

The AI Act: What Every Software Vendor Needs to Know

The European AI regulation is now in force. It imposes unprecedented obligations on software vendors integrating AI into their products. Here is what you need to know to anticipate, comply and turn this constraint into a competitive advantage.

Introduction

What Is the AI Act?

The AI Act (European AI Regulation) is the world's first comprehensive AI regulation. Adopted in 2024, it takes effect progressively between 2025 and 2027.

Its founding principle: regulate AI by risk level, not by technology. The more risk an AI system poses to fundamental rights, the stricter the obligations.

For software vendors, this means every AI feature must be assessed, classified and documented according to its risk level. Ignorance will not be an excuse.

Regulatory Framework

Risk Classification

Prohibited

Unacceptable Risk

  • Subliminal manipulation
  • Social scoring
  • Real-time biometric surveillance

Strict Obligations

High Risk

  • Employment and recruitment
  • Education and training
  • Credit access
  • Healthcare

Transparency Required

Limited Risk

  • Chatbots
  • Deepfakes
  • Inform user it is AI

No Specific Obligations

Minimal Risk

  • Spam filters
  • Video games
  • Content recommendations

Sector Impact

Which Software Is Affected?

High Risk

ATS / HRIS

Recruitment, employment access. Candidate scoring, CV matching, automated pre-screening.

High Risk (certain modules)

ERP

When used for HR management, payroll, financing. Affected modules fall under the high-risk category.

High Risk

EdTech / LMS

Training admission, automated grading. Decisions impacting access to education.

High Risk

FinTech / Insurance

Credit scoring, risk assessment, fraud detection. Access to essential financial services.

High Risk

HealthTech

Diagnostic support, medical classification. AI medical devices subject to enhanced requirements.

Limited Risk

CRM / MarTech

Recommendation, personalization. Transparency required: inform users of AI usage.

Requirements

Obligations for High-Risk Systems

Explainability

Every AI recommendation must be justifiable. The user must understand why AI made that choice.

Human Oversight

A human must be able to supervise, correct and disable the AI system at any time.

Complete Traceability

Every AI decision must be logged: inputs, outputs, model used, version, timestamp.

Bias Detection and Reduction

Regular testing, audited training data, mechanisms to correct identified biases.

Conformity Assessment

Complete technical documentation, robustness testing, audit before market launch.

Transparency

Inform users they are interacting with an AI system and how their data is processed.

Timeline

Application Timeline

February 2025

Ban on unacceptable risk systems

August 2025

Obligations for general-purpose AI models (GPAI)

August 2026

Full obligations for high-risk systems

August 2027

Full enforcement, penalties apply

"Ethics is not an optional module: it is the structural framework of the product."

Compliance by design costs less than retroactive compliance. Vendors who integrate AI Act requirements from inception avoid compliance debt and turn regulation into a competitive advantage.

Our Approach

How Rakam Helps

SafeBox

Proprietary bias removal and anonymization technology. Natively integrated into all Rakam systems to ensure fairness of AI decisions.

Compliant-by-Design

Compliant architecture from inception. No retrofit, no compliance debt. AI Act requirements are integrated at the product's core.

Sovereignty

Containerized systems deployable on-premise or sovereign cloud. Your data never leaves your servers.

Traceability

Complete logging, decision explainability, audit trail for every AI action. Every recommendation is justifiable and verifiable.

Sources

Regulation text: artificialintelligenceact.eu

European Commission: digital-strategy.ec.europa.eu

Jean de Bodinat, Polytechnique Insights — "Responsible AI is not optional"

Prepare Your AI Act Compliance

Free audit of your AI systems: risk classification, gap analysis and compliance roadmap.

Book a Call

Anticipate Regulation, Not Penalties

Leave us your details, we will get back to you within 24 hours.