AI in Accounting: Who Is Responsible for AI Errors?

The use of AI in accounting can accelerate processes, but it does not transfer liability to the software. In the event of an error, legal responsibility remains with the people who prepare, verify, approve or record accounting documents, as well as with those to whom the law assigns responsibility for organising the accounting function.

Thank you for reading this post, don't forget to subscribe!

Who Is Responsible When AI Makes a Mistake in Accounting?

Romania’s Accounting Law No. 82/1991 establishes the liability of the individuals involved in the accounting document workflow and, depending on how the accounting function is organised, that of the company’s administrator or the accounting service provider.

Therefore, the fact that an application has performed a classification or calculation does not mean that responsibility for the decision has been transferred to the software. An AI tool can support an accountant’s work, but it does not replace professional verification.

What Can Be Automated and What Remains the Accountant’s Responsibility?

Automation is safer for operations that can be objectively verified, such as processing invoices, extracting data from documents, archiving, sending notifications or querying public registers.

By contrast, tax classification, interpreting unclear situations, verifying VAT treatment and making approval decisions require professional judgment.

A simple criterion is traceability: every automatically generated result should be traceable to its source document and verifiable by a specialist. SimpluSPV can also be used to verify information available in public registers.

How Should an AI Tool Be Assessed?

Before implementation, a company should verify where data is stored, what happens when the system is uncertain, whether an audit trail of operations is available and what exactly the provider guarantees under the contract.

AI Literacy Is Already a Requirement

Article 4 of Regulation (EU) 2024/1689 requires providers and deployers to take measures to ensure a sufficient level of AI literacy, with this obligation applying from 2 February 2025.

In conclusion, AI can reduce repetitive work in accounting, but professional responsibility remains with people. The real benefit of automation comes when technology takes over verifiable tasks while the specialist focuses on analysis, control and decision-making.