China May Lock Down Its Best AI Models
Summary
Hosts Jason Howell and Jeff Jarvis discuss Beijing’s move to restrict access to its top AI models, mirroring U.S. export controls. They analyze OpenAI’s reported offer of a 5% equity stake to the U.S. government to secure GPT-5.6 approval, raising concerns about political influence. The episode also covers Meta using Instagram photos for AI training and Anthropic expanding Claude Co-Work to web and mobile platforms.
Topics discussed
Intro ads for Progressive and Netflix's The Diplomat
China's AI restrictions, talent drain, and NVIDIA chips
Anthropic's Coder tool and accusations of distillation
Anthropic's Claude Co-Work and Google's Spark update
Testing AI agents with Fable and benchmarking challenges
Sponsor reads for MailButler and Decagon
OpenAI's 5% equity offer to the US government
Gemini 2.5 launch and AI-driven ransomware attacks
Meta's Muse Engine and Ray-Ban Meta glasses privacy
Nothing Phone recording features and AI in schools
Sponsor reads for Gusto and Cisco Outshift
NVIDIA Rubin chip delays and Cloudflare's AI defaults
SpaceX AI rebrand, AI actor Norwood, and Reed Hastings
Outro, credits, and Working Smarter podcast promo
Listen ad-free on Castria