Dr. Philipp Mels Highlights AI Versus Copyright as Defining Legal Challenge of the 21st Century

In a detailed discussion with Leaders League, Dr. Philipp Mels identified intellectual property rights, trade secret protection, and data protection as the most critical legal areas to address beyond the AI Act when developing, operating, and using artificial intelligence systems.

Mels advises organizations to consider three pivotal stages in their AI lifecycle. The first involves acquiring, licensing, or developing AI systems, where copyrights related to the AI software and any embedded trade secrets must be reviewed, with patents playing a minor role in exceptional cases. The second stage concerns feeding and training AI with data, necessitating scrutiny of whether the data contains copyrighted content or trade secrets and whether appropriate confidentiality measures are in place.

The third stage pertains to the AI-generated results themselves, which may be protected under trade secret law or, in rare instances, copyright or design protection. Additionally, the use of these AI outputs must be carefully evaluated to avoid infringing third-party copyrights, design rights, trademarks, patents, or trade secrets.

Addressing the permissibility of using copyrighted or trade secret-protected works for AI training, Mels confirms that such use is possible only with the rights holder’s consent or license. Alternatively, publicly accessible sources may be utilized, provided they are properly identified. Under German copyright law, exceptions such as text and data mining permit data collection solely for AI training unless the copyright holder has issued a machine-readable copyright notice.

Mels highlights a prevalent practical risk: employees inadvertently inputting trade secrets into AI systems. This behavior can undermine a company’s trade secret protections by demonstrating a failure to maintain necessary confidentiality safeguards.

On the broader conflict landscape, Mels characterizes the most significant legal battle of the 21st century as one between AI providers and operators on one side and copyright holders on the other, who seek to enforce their rights against unauthorized AI use.

Data protection law also presents a critical challenge. The coexistence of the AI Regulation and the General Data Protection Regulation (GDPR) means that AI users must fully comply with GDPR requirements, including establishing a legal basis for data processing. The AI Act itself does not provide additional guidance on data protection compliance.

To navigate these complexities, Mels recommends that companies adopt a firm commitment to AI governance by developing binding AI policies. He advocates forming interdisciplinary AI teams that include legal, IT, management, and data protection officers to oversee AI system acquisition and usage. Regular audits and documentation of AI activities are essential, as is comprehensive employee training on AI capabilities and compliance with all relevant legal frameworks, including copyright law.

Dr. Mels’ insights underscore the intricate legal environment surrounding AI and the necessity for proactive, structured approaches to intellectual property, trade secrets, and data privacy to mitigate risks and foster responsible AI innovation.