#236 - GPT 5.4, Gemini 3.1 Flash Lite, Supply Chain Risk
Summary
This episode covers OpenAI’s GPT 5.4 release, highlighting its improved benchmarks, native computer use, and cyber safety protocols, alongside Google’s faster Gemini 3.1 Flash Lite. The hosts analyze the fallout from OpenAI’s Department of Defense contract, discussing Dario Amodei’s critical memo, the "cancel ChatGPT" trend, and rising ethical concerns. They also explore Luma’s new multimodal agents for creative work and debate the risks of AI agents causing real-world damage.
Topics discussed
Sponsors and Intro
Welcome and Episode Overview
OpenAI GPT 5.4 Release and Capabilities
OpenAI GPT 5.2 Instant and Hallucinations
Google Gemini 2.5 Flash Lite and Speed
AI Safety, Risks, and Luma Agents
Anthropic CEO Memo and OpenAI Controversy
Anthropic's Military Deal and Backlash
OpenAI's $110B Funding Round
Alibaba Qwen Team Departures
Policy: Anthropic Supply Chain Risk Designation
Lawsuit: AI Emotional Dependency and Safety
Anthropic Report on White-Collar Job Automation
Meta Opus 4.6 Evals and Outro
Listen ad-free on Castria