Most creators say they want protection from AI, yet a recent UK survey found over 90% would not opt out of AI access to their work. Why? Today’s visibility depends on online discovery and search, and those systems are powered by AI. If you opt out, your work may be safe from AI training, but it also risks becoming invisible. That dilemma, and our solutions, lie at the heart of my talk at Creative Australia.

I walked through through, at a high level suitable for a general audience, many of the main protection mechanisms across the creative–AI lifecycle, such as watermarking, fingerprinting, provenance tagging, unlearning and unlearnable work, and integrity detection. Each technique offers partial protection but also trade-offs in discoverability, privacy, and creative flexibility. Some safeguards can be bypassed cheaply, but still act as deterrents by leaving forensic traces when removed. The key is not choosing one fix but designing combinations that balance technical feasibility, effectiveness, and creative freedom.

Scientists, creators, government and industry can work together to rigorously test how well each method works, what are the trade-offs, and how these systems can reflect the values of creators and the broader Australian community. At CSIRO’s Data61, we’re committed to ensuring techniques are rigorously evaluated, continually improved, and that 𝘀𝘁𝗮𝗸𝗲𝗵𝗼𝗹𝗱𝗲𝗿-𝗱𝗲𝗰𝗶𝗱𝗲𝗱 trade-offs are visible and built into system design from the start.

𝗧𝗶𝘁𝗹𝗲: 𝗡𝗼 𝗦𝗶𝗻𝗴𝗹𝗲 𝗙𝗶𝘅: 𝗨𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱𝗶𝗻𝗴 𝗖𝗼𝗹𝗹𝗲𝗰𝘁𝗶𝘃𝗲 𝗣𝗿𝗼𝘁𝗲𝗰𝘁𝗶𝗼𝗻 𝗳𝗼𝗿 𝗖𝗿𝗲𝗮𝘁𝗶𝘃𝗲 𝗪𝗼𝗿𝗸𝘀
𝗛𝗼𝘀𝘁: Nicola Grayson, Head of Public Affairs, Creative Australia
𝗚𝘂𝗲𝘀𝘁 𝗦𝗽𝗲𝗮𝗸𝗲𝗿: Dr Liming Zhu, CSIRO’s Data61
𝗗𝗮𝘁𝗲: 𝗧𝗵𝘂𝗿𝘀𝗱𝗮𝘆 𝟮𝟬 𝗡𝗼𝘃𝗲𝗺𝗯𝗲𝗿 𝟮𝟬𝟮𝟱
𝗧𝗶𝗺𝗲: 𝟭𝟮-𝟭𝗽𝗺 (𝗔𝗘𝗗𝗧)

Slides: https://www.dropbox.com/scl/fi/w4rtl9kcrssruthmrd7o4/20251120-Creative-Australia.pdf?rlkey=na6s16bvgl7wl3ngf608rze9i&dl=0

By Sarah Firth – Original here https://www.linkedin.com/posts/sarahthefirth_graphicrecording-visualnotes-illustratednotes-activity-7397111216458014720-anpa