#181 - Google Chatbots, Cerebras vs Nvidia, AI Doom, ElevenLabs Controversy
Last Week in AI
00:00
Trending Models and Security Challenges
This chapter examines the sharp increase in downloads of Meta's Llama models, indicating potential use by cloud providers and raising security concerns in the context of AI technology in different global regions. The discussion also covers innovative approaches to training neural networks, focusing on the efficiency of new optimizers and their implications for distributed training. Lastly, it highlights increasing vulnerabilities in AI systems as recent findings reveal the effectiveness of multi-turn interactions in red teaming efforts, showcasing a pressing need for enhanced security measures.
Transcript
Play full episode
