Binder Grösswang, together with the Public Relations Association Austria (PRVA), hosted a panel discussion entitled “Artificial Intelligence and Copyright: Challenges and Opportunities for Marketing and Communication” at the law firm's office in Vienna.
On the panel: Ivo Rungg and Michael Horak, both partners in the Intellectual Property and Information Technology & Digital Law practice groups at Binder Grösswang. The discussion was moderated by Oliver Heiss, Managing Director of SAYO Brand Advisors and member of the PRVA Executive Board.
The discussion focused on the opportunities, risks, and legal implications of AI in marketing and communications. It became clear that clear and reliable regulations are essential in practice. Marketing and communications managers in particular need guidance in order to avoid legal pitfalls and use AI safely and responsibly. The topic is highly topical, exciting, and of crucial importance for practitioners—not least because AI applications have long since found their way into everyday business life and legislators are still catching up with developments in this area.
“We need to keep an eye on three key phases: training, prompting, and output. Copyright infringements can occur in all three phases – whether through the type of training, the use and input of protected data, or results that reproduce existing works,” explained Ivo Rungg at the outset to illustrate where copyright issues can arise when using AI.
Michael Horak followed up on this and explained the basics: “A copyright-protected work always exists when a person produces an individual intellectual achievement. In the case of AI output, this human contribution can lie in the formulation of prompts. Works are protected against any reproduction or adaptation that takes place without consent.”
The training methods were also a focus. The question of how models may be trained with regard to copyright law is of particular importance. While text and data mining are legally permissible under certain conditions, other forms of data collection fall into a gray area. It makes a significant difference whether a model is based on freely usable data or protected content – and which legal exceptions apply. Anyone who wants to prevent their own online data from being used for training purposes must clearly label it, for example with a machine-readable usage restriction.
It also became clear that problems can arise with both the input and output of AI systems. For example, if images from the internet are used without being checked or if generated texts are similar to existing works, copyrights may be infringed. International cases have shown that this can have significant legal and economic consequences.
Another focus was on the issue of liability. Depending on the contractual framework, various parties – from developers to users – may bear responsibility. For companies, this means clearly identifying potential risks and securing them at an early stage. “There is a growing number of generative AI offerings that can be used for a fee, are trained in a legally compliant manner, and for which the providers are also willing to assume liability,” says Ivo Rungg.
Particular attention must be paid to the assignment of rights for AI products. In principle, any rights belong to the person who creates the work as the author. However, if it is generated solely by AI, there is no copyright. In addition, employers must generally have the rights to their employees' works contractually assigned to them. In practice, this results in a high demand for internal guidelines and agreements to create transparency and legal certainty. “In principle, copyright law is well suited to the digital developments of our time, even if it should be adapted in detail and expanded to include regulations on investment protection in favor of the AI industry. Restricting protection to purely human works is no longer appropriate,” adds Michael Horak.
In addition to these issues, the possibility of protecting AI products was also discussed. Differentiated prompting, conscious revision, and clear documentation can help ensure that one's own creative work remains recognizable and is thus legally protected.
Finally, the two partners offered some practical tips:
“Be careful when training, be careful when setting prompts, and never forget to secure your own copyrights through documentation,” said Ivo Rungg and Michael Horak in unison.
Conclusion of the evening: Artificial intelligence is fundamentally changing the communications and marketing industry. Anyone who uses AI must consistently consider copyright law – otherwise, they face not only legal risks but also risks to their reputation. At the same time, the technology opens up enormous opportunities for creative processes and new forms of communication. Companies that develop clear strategies early on, create internal regulations, and continuously expand their knowledge can use the potential of AI responsibly and safely and position themselves for the future.