AI Audit Services
An independent review of the AI systems you already run: do they work, are they fair and secure, and do they meet the EU AI Act?
When you need an AI audit
An AI audit reviews AI systems that are already running. Companies order one before a regulatory deadline, after an incident, before selling a product into a regulated market or when a customer asks for evidence in procurement. It answers a simple question: does this system do what it should, safely and within the rules?
What we audit
- Performance: accuracy and stability on current data, not only on the original test set.
- Bias and fairness: whether results differ unfairly between groups.
- Security: prompt injection, data leakage and access control for LLM systems.
- Data protection: lawful basis, data flows, retention.
- Documentation and oversight: what the EU AI Act requires for the system's risk class.
How the audit works
- Scoping: we agree which systems and which questions.
- Evidence: documentation review, interviews and technical tests on your data.
- Findings: each issue with severity, evidence and a recommended fix.
- Follow-up: an optional re-test after fixes are in place.
AI audit or readiness assessment?
An AI audit looks at systems you already run. An AI readiness assessment looks forward and decides what to build next. Companies with AI in production often need both.
Frequently asked questions
How long does an AI audit take?
Two to four weeks for one system, depending on its complexity and the available documentation.
Do you audit third-party AI tools?
Yes, as far as the vendor's documentation and your configuration allow. We focus on how the tool is used in your company.
Is the audit enough for EU AI Act compliance?
It shows the gaps. High-risk systems also need ongoing risk management and a conformity process, which we can support.
Evidence that your AI works as intended
Tell us which systems are in scope. We will propose the audit questions, the tests and a timeline.