Liability for AI-induced errors in a company
These days, very many companies use artificial intelligence (AI) for offers, contracts, advice and internal decisions. If an AI system provides incorrect information or inappropriate recommendations, then the company generally remains responsible towards its customers. The crucial factor is the declaration the company has made or the service it has provided and whether it has fulfilled its duties of review and care.
Responsibility for AI results
From a legal perspective, AI is an auxiliary tool. If its results are accepted by a company, then these are usually treated as their own statements. Incorrect information may therefore trigger claims for rectification, price reductions, or damages. Approval by employees is not sufficient if they lack the relevant expertise to evaluate the results. This will also be true for offers and contracts.
Example: If an AI-generated offer contains an incorrect price and if a customer accepts it, then the company may be bound by it. It can only be contested under certain conditions and may give rise to claims for damages by the contractual partner.
Prior to sending, therefore, prices , services, deadlines, and liability clauses should be checked. The company likewise remains responsible for advice, calculations, forecasts and product recommendations. AI can support specialist knowledge, it is however not able to replace the check performed by a qualified person.
Claims against the AI provider
If a company has provided compensation for damages because of an AI error, then it can look into potential claims against the provider. However, in the case of general AI services, seeking recourse is frequently difficult because these usually do not come with a guarantee that the results will be accurate or fit for purpose. Moreover, proof will have to be provided of the specific system error.
Requirements under the EU AI Act
The EU AI Act classifies according to the risk associated with use. Since 2.2.2025 already, companies must ensure that their employees have sufficient AI knowledge. From 2.8.2026, further transparency obligations will apply. Under certain conditions, users must then be able to recognise that they are interacting with an AI system or that content has been generated artificially.
Stricter requirements apply to high-risk AI, such as for the selection of job applicants, personnel evaluations, or other significant decisions regarding employees. This may include human oversight, documentation, monitoring, and risk management. In addition, data protection, employment law, the protection of trade secrets, and contractual obligations will continue to apply.
Clear rules reduce the risk
Companies should exactly specify the tasks for which AI may be used, who will check the results and when an additional approval will be required. A competent person should always check, in particular, contracts, financial or legal information, personnel decisions and security-related planning. Important processes should be documented and confidential data should not be entered into public AI services in an uncontrolled manner. Insurance cover should likewise be reviewed because AI-induced damages are not automatically covered.
This may interest you
