#244 - GPT-5.5 Instant, Grok 4.3, OpenAI vs Musk
Summary
Hosts Andrei Kurenkov and Jeremy Harris discuss the Mythos AI security tool, defending its impact against skepticism regarding AI's cyber capabilities. They review OpenAI’s GPT-5.5 Instant, noting its improved coding skills and a quirky "goblin" training anomaly. The episode also covers XAI’s cost-effective Grok 4.3, Mistral’s new dense model, and Anthropic’s updated agent features like "Dreaming" for self-improvement.
Topics discussed
Intro, sponsors, and listener comments on Mythos
Discussion on Mythos, biosecurity, and AI safety
GPT-5.5 Instant, cyber capabilities, and the Goblin incident
xAI Grok 4.3 launch and pricing strategy
Mistral Medium 3.5, Anthropic's new agent features
Eleven Labs revamps AI music generation
Elon Musk's testimony in the OpenAI lawsuit
SpaceX and Anthropic compute partnership
Anthropic valuation, NVIDIA earnings, and data centers
Data center financing risks and DeepSeek funding
Anthropic's Natural Language Autoencoders research
US Government AI deals, NSA, and export controls
Deep dive: Introspection adapters for AI safety
Deep dive: Recursive agents and chain of thought
GPT-5.4 masters Connect 4 autonomously
Listen ad-free on Castria