🖤 'Poison' AI Training Data

🖤 Nightshade Tool Allows Artist to 'Poison' AI Training Data

Nightshade, a new tool developed by creators aiming to empower artists, allows users to “poison” the training data used by AI models.

This tool can corrupt the data used by AI art platforms like DALL-E, Stable Diffusion, and Midjourney, thereby disrupting their ability to generate images.

Nightshade introduces invisible changes to pixels in digital art, which confuses the model during training.

Ben Zhao, a professor at the University of Chicago and one of Nightshade’s creators, hopes this tool will counterbalance AI companies that use copyrighted data for training their models.

Artists can use Nightshade in conjunction with Glaze, another tool that masks their art style.

🎶 Spotify Is Making Major Changes to Its Royalty Model

Spotify is reportedly planning to overhaul its royalty model by 2024, which could significantly impact who gets paid for music streaming on the platform.

Spotify will fine music distributors for fraudulent activity and increase the minimum play time for noise tracks (background noise like rain, static, etc.) before they earn royalties.

Currently, any track that gets played for more than 30 seconds starts earning royalties, and many noise creators have gamed the system by breaking the noise up into short (even 31-second-short) tracks.

The biggest reported change is a minimum threshold of annual streams for a track before it starts earning royalties. A track would have to earn 5 cents per month in order to be paid, or about 200 streams per year.

Many indie tracks don’t hit this threshold, and so the pennies those artists would otherwise earn will be diverted to Spotify’s “streamshare” pool. When multiplied over the many, many low-play tracks on the streamer, it accounts for tens of millions of dollars.

That would instead get distributed to larger artists, who’d get a bigger share of the pooled money.

📸 Amazon Now Lets Advertisers Use Generative AI to Pretty Up Their Product Shots

Amazon is beta testing AI image generation tools for advertisers, aiming to enhance product images and improve ad performance.

The tool, which requires no technical expertise, allows brands to create lifestyle imagery around their products, leading to a potential 40% increase in click-through rates.

Despite some AI-generated artifacts in the images, the backdrops are generally convincing and align with Amazon’s existing product stock images.

Amazon CEO Andy Jassy has emphasized that generative AI is a significant focus for the company, indicating its potential to transform various aspects of its operations.

⚡ 3 Cool Tools

🧠 What did you think of today's email?

Your feedback helps me create better emails for you! Please reply to this email and let me know what you think.

Thanks for reading.

Until next time!

— Tay

p.s. if you want to sign up for the Rapid Rundown newsletter or share it with a friend or colleague, you can find us here.