5. What are the essential contractual clauses in the acquisition or integration of AI technologies
Every agreement with a provider of AI solutions (algorithms, software, generative models, etc.) should include a set of clauses aimed at protecting the operators acquiring a licence to use such technologies.
In particular, recent practice suggests paying specific attention to the following aspects:
- Intellectual property and rights over AI outcomes
- One of the most complex issues concerns the ownership of intellectual property rights over content created by AI systems.
In the fashion and luxury sector, for example, questions arise as to who owns the 3D models generated for virtual fashion shows, textile patterns or digital designs, advertising images created using generative tools, or texts and claims used in marketing and e-commerce. - In principle, as mentioned above, the recently enacted legislation does not confer any legal personality on AI systems or their providers, while it recognises copyright only in respect of creative works of human intellect, belonging to literature, music, figurative arts, architecture, theatre and cinematography, regardless of the form or mode of expression, even where created with the assistance of artificial intelligence tools, provided they constitute the result of the author’s intellectual work (Article 1 of Law No. 633/1941, as amended by Law No. 132/2025, in force since 10 October 2025).
- It is therefore essential to regulate contractually the ownership of intellectual works, in whatever form or manner they are created: for example, the agreement should specify that all works, outputs, models or improvements generated through a given AI system are the exclusive property of the interested company, or at least granted to such company under an unlimited and exclusive licence of use.
- Another crucial aspect relating to the ownership of AI-generated works and the improper use of confidential corporate information concerns the protection of company know-how. It is therefore essential to provide, for instance, that corporate data used or generated by AI systems and protected by industrial property rights may not be reused by the provider for other clients nor used to further “train” its AI model.
- One of the most complex issues concerns the ownership of intellectual property rights over content created by AI systems.
- Data protection and information security
- In sectors that process personal data relating to customers’ interests, habits and behaviour – particularly in fashion and luxury – AI systems often handle highly sensitive data. This includes, for example, the processing of VIP clients’ purchasing habits or the collection of biometric data for tailor-made digital applications.
In this context, as regards personal data, it is first of all essential to define privacy roles: the AI system provider may act as a data processor under the GDPR, subject to the execution of a specific Data Processing Agreement. Moreover, following a Data Protection Impact Assessment (DPIA), adequate technical and organisational measures must be implemented (such as encryption tools, data segregation and controlled access mechanisms), together with a prohibition of unauthorised or discriminatory profiling of the company’s customers. - Another key issue to be regulated contractually concerns rights of access, portability and reuse of data processed through AI systems, in order to avoid situations of technical or economic dependency on the supplier.
- In sectors that process personal data relating to customers’ interests, habits and behaviour – particularly in fashion and luxury – AI systems often handle highly sensitive data. This includes, for example, the processing of VIP clients’ purchasing habits or the collection of biometric data for tailor-made digital applications.
- Liability, compliance, warranties and audit rights
- It is undisputed that improper management of AI systems may generate erroneous, discriminatory or brand-damaging outcomes.
As previously discussed, the AI Act introduces a number of compliance obligations, including risk assessment and mitigation, data quality requirements, detailed technical documentation, registration in a European public database, human oversight, accuracy, robustness, cybersecurity and transparency towards users. - An appropriate contract should therefore include provisions protecting users, regulating the provider’s obligations, such as declarations of compliance with the AI Act and other relevant sector regulations, warranties regarding proper functioning, security and continuous updating of the system, indemnities and hold harmless clauses for damages arising from algorithmic bias, malfunctions or violations of third-party rights (e.g. copyright or personal data), and an obligation to promptly notify any incidents or vulnerabilities.
- Furthermore, to ensure proper control and transparency over AI system functionalities, it may be advisable to include contractual clauses granting technical and legal audit rights, also through specialised third parties, with an obligation on the provider to supply adequate information and documentation on the system’s operating logic.
- It is undisputed that improper management of AI systems may generate erroneous, discriminatory or brand-damaging outcomes.
- Ethical and reputational control
- The use of generative AI may result in outputs that are potentially offensive, stereotyped or culturally inappropriate, causing serious reputational damage. For this reason, it is advisable to include ethical AI usage clauses establishing obligations to comply with ethical principles, prohibitions on the use of data or images that infringe third-party rights or human dignity, and compliance with ESG and digital sustainability requirements, which are increasingly demanded by major luxury groups.