“At some point you have to wonder if this bug is a feature.”
— Alexios Mantzarlis, director of the Security, Trust and Security Initiative at Cornell Tech, reflects on xAI and Grok's penchant for identifying Elon Musk-friendly and/or far-right sources. Washington Post reports.
One more thing
The AI Lab wages a guerrilla war against exploitative AI
Back in 2022, the tech community was discussing image-generating AI models like OpenAI's Midjourney, Stable Diffusion, and DALL-E 2, which could follow simple verbal cues and depict fantasy lands or whimsical avocado chairs.
But the artists saw in this technological miracle a new kind of theft. They believed that the models were essentially stealing and replacing their work.
Ben Zhao, a computer security researcher at the University of Chicago, listened. He and his colleagues have created perhaps the most powerful weapons in an artist's arsenal against unauthorized image harvesting by artificial intelligence: two tools called Glaze and Nightshade, which add subtle distortions to image pixels so that machine learning models can't read them correctly.
But Zhao sees these tools as part of a battle to gradually shift the balance of power away from large corporations and back toward individual creators. Read the whole story.
—Melissa Heikkilä
We can still have good things
A place for comfort, fun and entertainment that will brighten your day. (Any ideas? Write to me or shoot skeet at me.)
+ If you ever feel like trying to recreate Painting by Jackson Pollockmaybe you should leave it to the kids.
+ Scientists have discovered that lions have not one, but two different types of roar 🦁
+ Relentless growth quarter zip must be stopped!
+ Wince: here's a brief history kisses 💋
