Ai Artificial Intelligence Archive

The White House asked AI companies to develop a watermark identifying AI-generated content. Some tech companies like Microsoft, Intel, and Adobe may have their answer in an internet protocol called C2PA, named after the Coalition for Content Provenance and Authenticity.

I just finished this deep dive from Jaime Brooks, aka Default Genders / Elite Gymnastics, on music technology’s march toward AI-synthesized voices like the the Drake / Ghostwriter situation, and why the current industry model — at least in the US — isn’t designed to support it.

DoorDash is testing a DashAI chatbot that’s supposed to make it easier to decide where to get your next meal, according to a report from Bloomberg. The chatbot will reportedly be able to provide suggestions about food and nearby restaurants based on your query.

Researchers from Carnegie Mellon University and the Center for AI Safety found that despite guardrails Google, OpenAI, and Anthropic built into chatbots, it’s still easy to get these to come up with dangerous answers. The researchers used a trick tested on open source chatbots that causes the system to bypass instructions preventing unfiltered results like asking ChatGPT to destroy humanity.

Humanity is fighting back after noticing that a series of low-quality websites appeared to be AI-scraping popular gaming subreddits to auto-generate articles. First it was the World of Warcraft subreddit posting about a completely made up character called “Glorbo,” now Destiny 2 players are getting in on the fun; baiting a website into writing complete nonsense.

On Meta’s Q2 earning call Wednesday, Mark Zuckerberg called Llama 2, the company’s latest generative AI model, an “open source project.”

Microsoft, Google, OpenAI, and Anthropic have teamed up to launch the Frontier Model Forum — a new industry body to promote responsible AI development.