AI Policy — How Aidos Uses Artificial Intelligence

AI Policy — Aidos

How we use artificial intelligence in our work

Version 1.0 — June 2026 | Reviewed annually, or when significant regulatory developments occur

Artificial intelligence tools are increasingly present in professional services — in document drafting, data analysis, research, and client communication. At Aidos, we use selected AI tools as part of our working practice. This policy explains our approach: what we use AI for, where we draw clear limits, and how we ensure that our professional responsibilities remain fully in human hands.

This policy applies to all Aidos team members and contractors involved in client work, and to all AI-assisted processes used in the delivery of our services. It is reviewed annually and updated as the regulatory and technological landscape evolves.

We use AI as a working tool, not as an adviser

AI tools can improve the speed and consistency of certain tasks: drafting internal documents, summarising public regulatory updates, structuring communications, and supporting research. We use them for these purposes.

We do not use AI to generate tax advice, accounting conclusions, legal assessments, or any output delivered to a client as professional guidance without full human review and sign-off. Every client-facing deliverable — whether a tax filing, advisory memo, payroll output, or compliance report — is reviewed, verified, and approved by a qualified professional at Aidos before it leaves our office.

Where AI tools are used in our workflows, they are commercial tools operating under enterprise-grade data agreements that include appropriate confidentiality and data processing controls. We do not use general-purpose consumer AI tools for client-adjacent work.

The professional judgement, accountability, and legal responsibility associated with our work remain with our team. AI does not alter that.

What we do not use AI for

We are explicit about our limits. We do not use AI to make or automate decisions affecting clients; we do not input confidential client data into AI tools outside of GDPR-compliant, enterprise-controlled environments; we do not publish or deliver AI-generated content as professional advice without human review; and we do not use AI systems that process personal data without appropriate data processing agreements in place. Where a tool or workflow does not meet these standards, we do not use it.

Data and confidentiality

Client confidentiality is a professional obligation that predates AI by decades. Our approach to AI tools is consistent with that obligation.

We do not enter identifying client information — names, tax identification numbers, financial figures, or other personal or commercially sensitive data — into AI tools outside of appropriately controlled environments. Where AI tools are used in client-adjacent workflows, they operate on anonymised, generalised, or internally generated content only.

Our data handling practices are governed by GDPR and our GDPR Policy. Nothing in our use of AI creates an exception to those obligations.

The EU AI Act and our compliance position

The EU AI Act (Regulation (EU) 2024/1689) entered into force on 1 August 2024 and is progressively applicable across all EU member states, including Bulgaria. As a professional services firm that deploys AI tools in a working context, Aidos falls within the scope of the regulation as a deployer.

The AI tools we use in day-to-day work fall within the limited-risk and minimal-risk categories of the Act. We do not deploy high-risk AI systems as defined under Annex III of the Regulation. Where the classification of a tool is uncertain, we apply a precautionary standard: human review is mandatory, and the tool is not used for consequential outputs.

We monitor regulatory developments under the AI Act — including the formalisation of Bulgaria’s national enforcement architecture — and will update our practices accordingly as EU-level guidance and national implementation measures are issued.

Human oversight is not optional

We consider human oversight of AI outputs to be a professional requirement, not a preference. In a regulated advisory environment — where the accuracy of a tax position, a payroll calculation, or a compliance assessment carries legal and financial consequences for the client — there is no acceptable substitute for qualified professional review.

AI can assist our team. It does not replace the expertise, accountability, or professional judgement that our clients engage us for.

Questions and updates

If you have questions about how we use AI in work that affects you, you are welcome to raise them directly with your contact at Aidos.

This policy sits alongside our Code of Conduct and our AML & KYC Policy as part of our governance framework. Our broader governance approach is described on our Governance page.