You might not be familiar with Kuaishou, but this Chinese organization just strike a big milestone: It’s introduced the 1st at any time text-to-video clip generative AI design that is freely out there for the community to check.
The short-video platform, which has around 600 million lively customers, introduced the new device, identified as Kling, on June 6. Like OpenAI’s Sora product, Kling is in a position to deliver films up to two minutes extensive from prompts.
But not like Sora, which nonetheless stays inaccessible to the community four months just after OpenAI debuted it, Kling has presently commenced permitting people try the design them selves. Zeyi Yang, our China reporter, has been placing it through its paces. Here’s what he built of it.
This tale is from China Report, our weekly newsletter masking tech in China. Indicator up to get it in your inbox each Tuesday.
Meta has produced a way to watermark AI-created speech
The news: Meta has made a technique that can embed concealed signals, regarded as watermarks, in AI-produced audio clips, which could aid in detecting AI-generated articles on line.
Why it issues: The tool, called AudioSeal, is the very first that can pinpoint which bits of audio in, for instance, a total hour-extended podcast may have been produced by AI. It could aid to tackle the growing dilemma of misinformation and cons using voice cloning resources. Go through the comprehensive tale.
—Melissa Heikkilä
