Claude Mythos, Project Glasswing and AI cybersecurity risks

Visit Mixture of Experts podcast page to get more AI content → https://ibm.biz/BdpKdN Is your infrastructure safe from the next generation of AI? This week on Mixture of Experts host Tim Hwang is joined by Kaoutar El Maghraoui and Martin Keen. First, Anthropic's Project Glasswing reveals why the company won't release its Mythos model andour experts discuss AI security. Next, OpenAI and Anthropic financials reveal divergent strategies—OpenAI leads consumer AI while Anthropic dominates enterprise, yet both face massive inference costs threatening profitability. Then, researchers tested whether AI can rediscover scientific breakthroughs by creating GPT-1900, trained only on pre-1900 knowledge, revealing that models excel at pattern recognition but struggle with paradigm-shifting discoveries. Finally, it’s Masters week! IBM Fellow Aaron Baughman demonstrates Masters Vault, an AI system that makes 58 years of golf tournament footage searchable through natural language queries. All that and more on today’s Mixture of Experts. 00:00 – Introduction 1:14– Anthropic Project Glasswing and Mythos model 16:17 – OpenAI and Anthropic financials revealed 26:10 – GPT-1900: Can AI rediscover scientific breakthroughs? 37:57– Masters Vault: AI-powered golf archive demo with Aaron Baughman The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. Subscribe for AI updates → https://ibm.biz/BdpKd7 #AnthropicGlasswing, #AIcybersecurity, #OpenAIfinancials