AI now touches every part of the creative lifecycle — how works are made, shared, discovered, and, at times, learned from by AI systems without permission.

This session explores how creators can strengthen protection across this cycle while still using AI to reduce repetitive tasks and enhance discovery.

Examples include techniques that make artworks impossible for AI to learn from, watermarking or provenance systems that record how and when works were created, and tools that detect whether content was made by a person or by AI — and which AI. No single safeguard is perfect; effectiveness comes from how these mechanisms work together.

Each protection involves trade-offs in cost, usability, removability, and in how easily its removal can be detected and attributed. Even when a protection can be easily removed, the ability to detect and attribute that removal can itself deter misuse.

CSIRO is inventing new protection mechanisms and studies these trade-offs scientifically to understand how each protection works, where it fails, and how they can complement one another. This research helps creators and policymakers identify the most effective mix of techniques to support a creative ecosystem that is fair, transparent, and responsibly innovative.

Watch the full presentation here

 

Hosted by:
Nicola Grayson, Head of Public Affairs, Creative Australia

Presenter:
Dr Liming Zhu
Research Director, Software and Computational Systems
DATA61 | CSIRO
Liming.Zhu@data61.csiro.au