Description

Send us a text

New rules now apply to general-purpose AI models, the very foundation of tools like ChatGPT and Gemini. Whether you’re training, deploying, or just integrating these models, your legal obligations have just changed. The new rules came into place as of August 2nd, 2025.

What’s Changed?
The  EU rules focus on three areas: transparency, copyright, and safety.

1. Transparency
   
   • Providers must disclose information about how the model was trained and designed.
     
   • That includes publishing a public summary of training data sources using a standard EU template.
     
     
2. Copyright Compliance
   
   • Any copyrighted content used in training must be properly documented.
     
   • This opens the door for rights-holders to make claims or demand compensation.
     
3. Technical Documentation
   
   • Detailed records about model training, function, and limitations are now mandatory.
     
4. Systemic-Risk Models
   
   • If your model exceeds 10²⁵ FLOP, that’s the EU’s risk threshold, you must notify the Commission and comply with additional safety and security rules.

Want to go deeper on AI?

📖 Buy AI Playbook

📩 Get my weekly LinkedIn newsletter, Human in the Loop.

🎓 Level up with the CPD Accredited AI Playbook Diploma

📞 Let's talk about AI training for your team: digitaltraining.ie or publicsectormarketingpros.com if you are in government or publics sector.