AI Safety Foundations
Master the fundamental principles of security, ethics and governance of AI systems to deploy responsible and compliant solutions.
Key skills
- ✓Identify and mitigate security risks in AI systems
- ✓Apply AI ethics and governance principles
- ✓Evaluate the robustness and reliability of ML models
- ✓Ensure regulatory compliance (AI Act, GDPR) of AI projects
The question bank for this exam is prepared after your purchase. The exam is available no later than 48 hours after payment; you will receive an email as soon as it opens.
Is this right for you?
✓ This certification is right for you if:
- →Data Scientists and ML Engineers
- →AI Managers and AI Product Managers
- →Consultants and auditors in AI governance
- →Risk Managers and DPOs working on AI systems
✗ This certification is not suitable if:
- →Beginners with no AI experience: first orient yourself towards fundamental machine learning training before addressing security.
- →AI Safety experts seeking an advanced certification: consult our Master 801 AI Risk & Governance Architect pathway to deepen your knowledge.
Prerequisites & eligibility
Online registration, payment by credit card or bank transfer, identity verification during the supervised exam.
Purchase options
Exam voucher, LMS pathway, bundle packs — build your order
Options d'achat
Composez votre parcours — voucher d'examen, préparation LMS, packs entreprise.
Skills assessed
Model Security
Detection and prevention of adversarial attacks, data poisoning and information leakage.
AI Governance
Implementation of governance frameworks aligned with the European AI Act and ISO standards.
Ethics and Explainability
Application of fairness, transparency and interpretability principles for algorithmic decisions.
Regulatory Compliance
Navigating the European regulatory landscape (AI Act, GDPR, NIS2) applied to AI.
Exam format
Curriculum
AI Safety Fundamentals
AI risk taxonomy according to NIST and EU AI Act. ML system attack surface (data pipeline, model, inference, infrastructure). Trustworthy AI principles (robustness, transparency, fairness, accountability). International standards: ISO/IEC 42001 (AI management system), ISO/IEC 23894 (risk management), NIST AI Risk Management Framework. Mapping of actors and regulations (AI Act, GDPR, sectoral).
Model Technical Security
Adversarial attacks: FGSM, PGD, C&W, black-box attacks. Data poisoning (poisoning attacks) and backdoors. Model inversion and model extraction. Membership inference attacks. Prompt injection and LLM jailbreaking. Defenses: adversarial training, certified defenses, anomaly detection. Generative model red teaming. Securing the MLOps pipeline. Secure management of artefacts and model registers.
Governance and Compliance
European AI Act: system classification (prohibited, high-risk, transparency, minimal). Obligations for high-risk systems: technical documentation, registers, continuous monitoring. Conformity assessment procedures and notified bodies. GDPR applied to AI: lawfulness of processing, profiling, automated decisions. DPIA (Data Protection Impact Assessment) for AI systems. Model registers and traceability. Algorithmic audits: methodologies and frameworks. Organisational governance: roles, AI committees, internal policies.
Ethics and Explainability
Identification and measurement of algorithmic bias: data bias, model bias, deployment bias. Fairness metrics: demographic parity, equalized odds, calibration. Bias mitigation techniques. Explainability (XAI): LIME, SHAP, Integrated Gradients, attention mechanisms. Global vs local interpretability. Counterfactual explanations. Human-in-the-loop and human oversight. Accountability and algorithmic responsibility. Assessment of the social and environmental impact of AI systems.
How to fund your certification
3 ways to cover the cost of your certification
OPCO — Employer funding
Ask your employer to cover the cost through your OPCO (employee training plan).
Personal — €450 incl. VAT
Secure online payment by credit card (Stripe).
Business rate: €585 excl. VAT
Business — custom quote
Teams, cohorts or private catalogue: a quote tailored to your volumes and goals.
Frequently asked questions
Does this certification cover the European AI Act?
Yes, the framework deeply integrates the requirements of the AI Act, including the classification of high-risk systems, documentation obligations and compliance procedures. You will be able to assess the risk level of an AI system and identify applicable regulatory obligations.
Do you need to be a developer to take this certification?
No, although a technical understanding of ML systems is required. The certification is aimed at both technical profiles (Data Scientists, ML Engineers) and governance profiles (Risk Managers, AI Product Owners).
Does the preparation material include practical labs?
Yes, the LMS pathway includes sandbox environments to test adversarial attacks on models, manipulate XAI tools (SHAP, LIME), and simulate AI Act compliance audits. You will work on anonymised real datasets.
Is this certification recognised by regulatory authorities?
PROVA AI Safety Foundations is aligned with ESCO and EQF frameworks recognised in the EU. Although not mandatory from a regulatory standpoint, it demonstrates to authorities (CNIL, sectoral authorities) a proactive approach to AI compliance and risk management, facilitating audits and sectoral certifications.
Your certification pathway
Before, during, after — the recommended progression
Droits du candidat
Transparence
Les critères de décision et le barème sont documentés et accessibles avant l'examen.
Droit d'appel
Toute décision peut être contestée dans les 30 jours. Examen par un tiers indépendant.
Plainte
Toute personne peut signaler un dysfonctionnement. Formulaire public accessible sans compte.
