AI AUDIT

AI and machine learning audit

We review your company's AI models, assistants and agents: security, personal data, bias and compliance, with a prioritized report.

Book a free consultation
— AI audit —

AI you can audit

An AI audit verifies that the artificial intelligence systems your company builds or uses work as intended, do not leak data, do not discriminate and comply with the rules that apply to you. T.I. RESCUE runs it with the same rigor as our cybersecurity and ISO audits: a clear scope, technical testing, evidence and a prioritized remediation plan.

— Scope —

What do we review?

AI inventory

Which models, assistants, agents and AI features the company uses, including shadow AI.

Security

Prompt injection, data leakage, over-privileged agents and API abuse, based on the OWASP Top 10 for LLM.

Personal data

Consent, purpose, cross-border transfers and vendors under applicable privacy law.

Bias and quality

Accuracy, fairness across groups, model drift, and data quality and provenance.

Vendors

AI service contracts: training on your data, where data is processed and responsibilities.

Governance

AI policy, owners, impact assessment and monitoring under ISO/IEC 42001.

— Deliverables —

What do you get?

Findings report

Each risk with its evidence, priority and the recommendation to close it.

Technical testing

Controlled tests against your assistants and agents: attempts to extract data, bypass rules or trigger unauthorized actions.

Remediation plan

What to fix first, who does it and how to verify it.

Verification

We re-test critical findings after you fix them.

— Methodology —

Frameworks

ISO/IEC 42001

The certifiable standard for AI management systems: requirements and 38 Annex A controls.

NIST AI RMF

The US AI risk management framework: govern, map, measure and manage.

OWASP Top 10 for LLM

The most common security risks in applications built on language models.

AI and privacy laws

The EU AI Act, state laws such as Colorado's, and privacy laws such as Colombia's Law 1581.

— Free —

Free AI governance tools

ISO 42001 self-assessment

22 requirements, score by domain.

Shadow AI checklist

Is your team using AI without control?

AI risk classifier

Is your AI high risk under the EU AI Act or US state laws?

AI use policy generator

Your AI acceptable use policy in 2 minutes.

AI audit frequently asked questions

What is an AI audit?

An independent assessment of a company's artificial intelligence systems: how they are governed, what data they use, how secure they are against attacks, and whether their results are accurate and fair.

Do you audit in-house machine learning models?

Yes. We review training data, validation, production monitoring and the security of the model and its API, plus the process that governs it.

What if we only use ChatGPT, Copilot or another assistant?

It still applies: most of the risk is in what data goes in, under which accounts and under what contract. Start with the shadow AI checklist.

Does it include penetration testing of the AI?

Yes, when it is in scope: we test your assistants and agents the way an attacker would. For the rest of your infrastructure there is our penetration testing.

Does it help with ISO 42001 certification?

The audit identifies the gaps against the standard. If you want to get ready for certification, see our ISO 42001 AI governance service.

How much does it cost?

It depends on how many AI systems are in scope and whether technical testing is included. Request a quote.

Do you know which AI your company uses, and with what data?

T.I. RESCUE is a Colombian cybersecurity company based in Medellín.

Contact us