In this episode of Decrypting AI, hosts Ryan Ozawa and Jason Nelson discuss recent developments in AI security and applications. They cover a study revealing vulnerabilities in AI models, with Grok having the worst security and Llama the best. The conversation also touches on Higgsfield, a company creating animated AI-generated social media content, and Microsoft's report on China's use of AI to spread disinformation and disrupt US elections. Lastly, they explore the latest in AI-generated audio, including tools like Stable Audio 2, Suno 3, and Udio, and discuss the challenges surrounding copyright and the AI rumor mill.
","author_name":"Ryan Ozawa and Jason Nelson"}