AI training
with Facebook and
Instagram.
AI training
with Facebook and
Instagram.
from
Can Meta train its artificial intelligence with the public data of Facebook and Instagram users? The Cologne Higher Regional Court has issued a landmark ruling on the use of publicly accessible data for AI.
Data for smart AI
The proceedings centered on the question of whether the operator of the globally renowned social media platforms Facebook and Instagram, Meta Platforms Ireland Limited, is allowed to use publicly accessible user data for the development and improvement of its AI systems.
A consumer association had applied for an injunction to prevent this use of data. The issue was therefore whether Meta was allowed to use the data that users share publicly anyway to make its AI “smarter” and better adapt it to regional characteristics.
The association argued that the use of publicly accessible user data for AI training constituted unauthorized processing of personal data. It feared that this could infringe users’ personal rights and violate the provisions of the General Data Protection Regulation (GDPR). In particular, the question arose as to whether users were sufficiently informed about the intended use of data and whether this use was proportionate.
Meta, on the other hand, argued that the development and improvement of AI systems is a legitimate business interest. Powerful AI could offer users better services and personalized experiences. Meta also emphasized that measures are being taken to minimize the impact on user data, for example through anonymization or pseudonymization where possible and by complying with other requirements of GDPR. The company also argued that the adaptation of AI to regional conditions is only possible by analyzing the relevant data.
Cologne Higher Regional Court gives green light with conditions
The OLG Cologne has ruled Judgment of 23.05.2025 – Ref. 15 UKl 2/25 rejected the application for a preliminary injunction. The court found that the social media company’s interest in training its AI, including adapting it to regional customs, was legitimate. It recognized that Meta has a legitimate interest in the innovation and further development of its products.
The decisive factor for the court was apparently that the measures taken by the operator to mitigate the impact on user data were deemed sufficient.
The use of data for AI training is permitted, provided that all other requirements of the General Data Protection Regulation are met.
This means that permission to use data is not a free pass, but remains linked to full compliance with strict European data protection regulations.
What does this ruling mean?
The ruling states that the development of artificial intelligence can be a legitimate business interest, even if publicly accessible data is used for this purpose. It represents a compromise between freedom of innovation and data protection , which opens the door for the further development of AI, but at the same time reaffirms the limits of GDPR.
For technology companies, this means that they can generally invoke a legitimate interest in training data with regard to personal data.
Critical view
Limits of “public”: Even if data is “public”, this does not necessarily mean that users consent to every type of processing. The question of consent and transparency therefore remains crucial.
Practical enforcement of the requirements: Compliance with “all other requirements of GDPR” is a high hurdle. It remains to be seen how effectively this will be monitored and enforced in practice. For example, how should particularly sensitive data in photos be handled? How can this be implemented and complied with in detail?
Data protection concerns therefore remain even after the ruling.
Conclusion
The judgment of the Higher Regional Court of Cologne is an important building block in the still young case law on artificial intelligence. It shows that courts are trying to find a balance between technological progress and the protection of privacy. It remains to be seen whether the Cologne Higher Regional Court’s reasoning will hold up in the long term.
Users who do not want to make their data available as training data should therefore reduce their content on social media accordingly.
For companies that develop AI, the ruling shows a way to process personal data in a legally compliant manner.
We are happy to
advise you about
AI!
