The Obtain: Video-generating AI, and Meta’s voice cloning watermarks
You might not be conversant in Kuaishou, however this Chinese language firm simply hit a serious milestone: It’s launched the primary ever text-to-video generative AI mannequin that’s freely obtainable for the general public to check.
The short-video platform, which has over 600 million energetic customers, introduced the brand new instrument, referred to as Kling, on June 6. Like OpenAI’s Sora mannequin, Kling is ready to generate movies as much as two minutes lengthy from prompts.
However not like Sora, which nonetheless stays inaccessible to the general public 4 months after OpenAI debuted it, Kling has already began letting folks strive the mannequin themselves. Zeyi Yang, our China reporter, has been placing it by way of its paces. Right here’s what he made from it.
This story is from China Report, our weekly e-newsletter protecting tech in China. Enroll to obtain it in your inbox each Tuesday.
Meta has created a strategy to watermark AI-generated speech
The information: Meta has created a system that may embed hidden indicators, referred to as watermarks, in AI-generated audio clips, which may assist in detecting AI-generated content material on-line.
Why it issues: The instrument, referred to as AudioSeal, is the primary that may pinpoint which bits of audio in, for instance, a full hour-long podcast might need been generated by AI. It may assist to deal with the rising drawback of misinformation and scams utilizing voice cloning instruments. Learn the complete story.
—Melissa Heikkilä
