American startup Tavus announced that its Griffin model was mistaken for a human by 48 % of participants during one-minute video calls, with 26 out of 54 people fooled, compared to only 2.4 % for the previous Phoenix-4.5 generation. Griffin simultaneously analyzes voice, image, facial expressions, pauses, and conversation rhythm. The model scores 3.83 out of 5 on Nvidia’s VideoFDB benchmark, close to the 3.92 recorded for human interactions. This breakthrough comes as deepfake frauds multiply, with spectacular cases involving amounts exceeding 90 million euros. Tavus, founded in 2020 in San Francisco, has raised approximately 64 million dollars to develop this technology.
Source: Read the original article
Disclaimer: this content is for information purposes only and is not financial advice. Cryptocurrencies are highly volatile: you may lose all of your capital. Always do your own research. Legal notice
