EU AI Act: Commission is consulting on guidelines and a code of practice for transparency of AI-generated content
14 September 25
Elisar Bashir
The public consultation aims to gather practical examples of AI systems that directly interact with natural persons, which are subject to transparency requirements under the AI Act. The EU executive body further seeks feedback on existing state-of-the-art technical solutions for marking and detecting AI-generated or manipulated content as well as the concept of deep fake generating AI systems.
EU’s top court rules on when pseudonymous data is to be considered personal data
11 September 25
Alessandra Vaes
The EU Court of Justice clarified that the classification of pseudonymous data as personal data depends on whether the recipient has the means to re-identify the concerned individuals. It thereby endorsed the relative nature of personal data in accordance with its previous rulings. The court also clarified that transparency and information requirements towards data subjects are incumbent upon the data controller irrespective of whether the pseudonymous data in the hands of the recipient is identifiable or not.
US: Judge orders behavioural remedies in Google Search antitrust case
09 September 25
Jose Jehuda Garcia
The judge presiding over the antitrust infringement case against Google Search rejected all manner of structural remedies (such as the forced divestiture of Google companies). However, the court ordered behavioural restrictions for Google to observe going forward. Both Google and the US Department of Justice signalled intent to appeal the court´s decision.
US plots bold new AI policy course with “US AI Action Plan”
19 August 25
Jose Jehuda Garcia
This Global Trends Flash discusses the new US AI Action Plan (AIAP). This plan outlines over 90 federal policy actions aimed at achieving global artificial intelligence (AI) “dominance” for the US. The AIAP amounts to a stark departure from the prior presidential administration’s risk-focused AI policy approach. In many jurisdictions around the world, AI risk management is also at the heart of AI policy.
EU AI Act: Commission issues mandatory template for summary of content used to train general-purpose AI models
30 July 25
Elisar Bashir
The template requires the listing of data sources used to train the model such as public and private datasets as well as measures taken to avoid or remove illegal content under EU law from the training data. It is obligatory for all providers of GPAI models, including those released under free and open-source licenses. Providers of GPAI models already available in Europe before 2 August 2025 will have two years to publish their summaries.
EU AI Act: Commission issues non-binding guidelines for providers of general-purpose AI models
30 July 25
Elisar Bashir
The guidelines give the European Commission’s interpretation of how the AI Act applies to providers of GPAI models. They establish a threshold for all GPAI models covered by the new EU rules and clarify the notification procedure for those models with systemic risks. They also address the effect of signing up to the recently published code of practice. The provisions on GPAI models enter into application on 2 August 2025.