AI Vendor & Model Due Diligence

Катерина Дубас Legal IT Group
Kateryna Dubas
Head of Privacy and Cybersecurity Practices
Send a request
We assess a third-party model provider, API or SaaS before integration: where your data is processed, whether the model is trained on it, which transfer mechanism applies, what their DPA says, and what residual risks remain. Deliverable: a due diligence memorandum with a clear go / no-go recommendation.
Кібербезпека Legal It Group

Why Us

Because a standard security questionnaire doesn't catch AI-specific risks

Certified privacy-engineering and cybersecurity specialists who check the provider's security and look for all available information on how the model was trained
  • A team of three FIPs

    Certified privacy professionals: CISSP, CIPP/E, CIPP/US, CIPM, CIPT, FIP.
    icon
  • We turn promises into verifiable facts

    We don't take "opt-out exists" on faith; we look for evidence of data isolation and other controls
    about_icon2
  • We know where a DPA's weak points are

    We read the DPA, subprocessors, transfer mechanisms, certificates and audit reports
    about_icon3
  • Legal + tech in one team

    We understand the integration architecture, not just the vendor's promises
    about_icon4
Send a request
decor
Кібербезпека та privacy engineering - Legal IT Grpup

We help teams make an informed decision about integrating a third-party model, with a clear understanding of where your data will go, what risks you are accepting, and under what conditions integration is possible at all


title_icon

Our Services

  • Provider and model assessment

    icon-vectors
    We analyze who you are integrating with: where your data is physically processed and stored, whether it is used to train the vendor's model, which subprocessors are involved, which certifications are confirmed (ISO 42001, ISO 27001, SOC 2) and whether they have expired.
  • Analysis of data, transfers and the DPA

    icon-vectors
    We review the legal and technical side of the data flow: whether there is a valid DPA, and what applies to data residency, the exit process and data deletion after termination of the agreement. Separately (if needed), we assess the model's output risks (bias, reliability, hallucinations).
  • Due diligence memorandum: go / no-go

    icon-vectors
    The deliverable is not a questionnaire but a memorandum listing identified risks grouped by criticality, with a clear recommendation: where the risk is acceptable and it's a go with conditions, and where it's a no-go, explaining what is missing.
  • Preparing documentation for integration

    icon-vectors
    If the decision is go, we bring the integration to readiness and update or edit everything the new model touches: the DPA with the vendor (review, editing, negotiating insufficient clauses), RoPA, privacy notice, DPIA for the new model, as well as internal AI policies and registers.


Certified Experts

Our team holds certifications in privacy and cybersecurity

Internationally Recognised Expertise

FIP_Dubas
Катерина Дубас_кібербезпека
FIP_Tarasiuk

blog_iconRelevant and Practical Articles

  • Data protection officer
  • AI compliance officer
  • Data privacy compliance
  • Diia.City
  • Digital Millennium copyright Act
  • Trademark in IT
Go to Blog

We Write About What We Practise

ІP, GDPR, contracts and disputes, as well as the legal aspects of implementing technologies such as artificial intelligence and best practices for implementing specific legal solutions.