The Ad Hoc AI Licensing Regime
Summary
This episode analyzes the emerging ad hoc US government licensing regime for frontier AI, detailing delays to GPT-5.6 and restricted Mythos access. The host critiques this opaque process as harmful to innovation while highlighting industry shifts toward open-source models like GLM-5.2 and CEO-led AI strategies. It also covers Micron’s strong earnings, OpenAI’s delayed IPO, and new integrations like Claude Tag in Slack.
Topics discussed
Intro and note on post-recording news updates
The Fable 5 saga and ad hoc licensing regime
GPT 5.6 delay and government approval process
Rise of open source models and Claude Tag integration
Enterprise shift to in-house models and CEO-led AI
Market signals: Micron earnings and OpenAI IPO
Mythos access lifted for select partners
GPT 5.6 limited preview details and closing thoughts
Listen ad-free on Castria