Generative AI and copyright – what every employer needs to know

Artificial intelligence has burst onto the business scene: it assists with writing e-mails and taking notes, creating products and services, and even designing advertising materials. The operational benefits for employers are undeniable, but they are accompanied by serious risks regarding copyright and other intellectual property rights. We identify quite a few in our practice. Below, we highlight two key areas that have become shrouded in certain myths and misconceptions.

“The employee generated this using AI, so the copyright belongs to the employer”

We regularly encounter the belief among employers that output generated by employees using AI tools – graphics, marketing texts, presentations, chunks of code, automatically constitute the employer’s intellectual property. The employer intuitively assumes that since the output was created during working hours, as part of their duties, copyright law (Section 12 of the Copyright Act) automatically grants them ownership. Unfortunately, this is not the case, because under copyright law, only a human-created work can be considered a work and be subject to copyright protection. Put simply, if an employee enters a prompt into an AI tool and then receives a finished result, such result is produced through algorithmic data processing, not through human creation. Since the output is not a work, no copyright arises in relation to it. There are, therefore, no rights that the employer could acquire. Interestingly, business owners are often surprised when we point out  that purchasing paid access to the AI tool does not alter the above. Such payment does not change the rules of copyright law.

What risks does this pose? Whilst generating an email or a simple presentation may in itself be of little consequence to an employer, the use of AI in a company’s key products (e.g. in software, in graphics on product packaging, or in materials for language learning or tutoring) may render that product unprotected. An output generated by AI is not protected – it can be freely used and copied by an employee, a competitor or any third party. Nor does it form part of the company’s assets as a business asset, and it cannot be effectively transferred. This may be crucial in investment transactions (e.g. the sale of a company) or when applying for grants.

This risk can be minimised. If an employee modifies the generated output, combines it with other elements, or adds their own structure, content or graphic elements, then such a work or its component, constituting an expression of individual creativity, may be subject to copyright protection. The principle is simple: the more creative work an employee puts in themselves, the greater the chance that the resulting material will be,  at least to some extent,  protected by copyright.

“AI did it – let’s post it on social media – we’ll get reach”

AI is excellent for quickly creating graphics, videos or advertising and promotional texts for a company’s social media. Unfortunately, such material sometimes infringes the intellectual property rights of third parties (if, e.g.  an AI-generated graphic resembles a well-known logo or a distinctive character associated with a competitor). Worse still, the employer usually only finds out about this when they receive a cease-and-desist notice from the rights holder,  by which time the material has already had a chance to circulate in the public domain. It is important to remember that, in accordance with general principles, the employer is liable for such infringing material.

It is true that AI tools are becoming increasingly effective at limiting the generation of output that infringes on others’ rights. However, it still happens that the output contains protected elements  e.g. those strikingly similar to registered trademarks, industrial designs or protected works,  or is so close to them that it poses a real risk of infringement. Crucially, an employee need not have any intention or even be aware that the material generated by AI uses protected content. For the purposes of assessing an infringement of intellectual property rights, intention or awareness are irrelevant.

Depending on the type of infringement and the nature of the infringed right, the rights holder may not only demand the cessation of further actions (e.g. removal of the material from social media), but may also bring a range of other claims: publication of an appropriate statement, removal of the effects of the infringement, and in more serious cases,  compensation or the surrender of unjustifiably gained benefits. The range of potential consequences is, therefore, wide, and the costs,  particularly those relating to reputation can be severe.

This raises the question of whether an employer can subsequently pursue claims against an employee who generated and disseminated such material. According to Articles 114–122 of the Labour Code, the answer is, in principle, yes. If the employee can only be held liable for unintentional fault, their liability is limited to three months’ salary. In the case of intentional fault, the employee is liable for the full amount of the damage. However, in each  case, the employer must prove the damage and a causal link to the employee’s actions.

A separate issue is whether the employee’s use of material infringing on the rights of others allows the employer to terminate their employment. If such conduct was deliberate (e.g. carried out in breach of internal procedures or despite having undergone relevant training), it may be regarded as a breach of fundamental employee duties, which justifies contract termination without notice (Section 52(1)(1) of the Labour Code). Where the employee cannot be accused of intentional fault or gross negligence, disciplinary action will be unjustified, but this does not preclude contract termination. Everything depends on the circumstances of the specific incident.

Ways to minimise risks – training and internal procedures

In light of the above, we recommend, first and foremost, educating staff – awareness of the risks is absolutely essential. It is worth bearing this in mind when designing internal processes and procedures for the use of AI. The aim is not to prohibit staff from using AI, but to establish clear guidelines.

For example: it is worth stipulating that the company’s key assets – products, services or materials made available to customers,  cannot be created solely using AI or must contain a significant creative contribution from an employee. It is equally important to define documentation rules: an employee should indicate which elements of the material constitute their own contribution and which were generated by AI.

From the organisation’s perspective, it is also crucial to implement specific procedures to verify AI-generated content before its publication and use. Such verification,  even at a basic level,  allows potential infringements to be identified before the material enters the public domain.

This is not bureaucracy – it is a safeguard that can determine whether the organisation holds the rights to the material in question at all, or whether there has been an infringement of third-party rights.

Summary

The use of generative AI within an organisation brings tangible benefits, but also generates legal risks. The two areas discussed above, the apparent copyright protection of AI outputs and the risk of infringing third-party intellectual property rights, are issues that an increasing number of companies are facing in practice, often only once the problem has already arisen. The key to minimising these risks is not to abandon AI tools,  which would be out of touch with market realities. What is essential is the conscious management of processes, which will help eliminate the most serious threats to the organisation.

This article is based on a presentation given during the HR Law 2026 webinar. Trends

Lena Marcinoska-Boulangé, adwokat, intellectual property practice