Security-rooted EU AI Act conformity assessments for Swiss organisations. Prompt injection audits, bias detection and systematic risk analysis.
From initial risk classification to a comprehensive security assessment — choose the package that fits your requirements.
Systematic classification of your AI systems according to EU AI Act risk levels. Including gap analysis and recommendations.
Complete conformity assessment per Art. 9–15 EU AI Act. Technical documentation, risk management and audit report.
Comprehensive AI security assessment: red teaming, prompt injection tests, data leakage analysis, bias detection and complete documentation.
Our testing framework covers the critical attack vectors and vulnerabilities of modern AI systems.
Systematic tests against direct and indirect prompt injection attacks. OWASP LLM Top 10 compliant.
Testing for unintended disclosure of training data, PII and confidential information.
Quantifying hallucination rates with domain-specific benchmarks and factual accuracy metrics.
Detection of systematic biases in model outputs. Fairness metrics across protected attributes.
Tests for access control circumvention, role escalation and sandbox escape in AI agents.
Validation of output safety: toxic content, unauthorized actions and compliance violations.
Researcher and security engineer with a background in AI safety and regulated Swiss infrastructure. Combines academic depth (Harvard, Cambridge) with hands-on experience building secure AI systems.
Let us assess your AI systems. The initial consultation is non-binding.
contact@ai-assurance.ch →Entry from CHF 3,000 · Conformity assessment from CHF 10,000 · Full assessment from CHF 25,000