Descriptions:
David Ondrej sits down with a guest to unpack what they describe as an unprecedented moment in AI history: for the first time ever, two major frontier models — Anthropic’s Fable and OpenAI’s GPT-5.6 — have been withheld from general public release within a matter of weeks, with the US government effectively directing the release cadence of the most capable AI systems in the world. GPT-5.6 launched in a limited preview available only to a small group of government-approved partner companies, with broader access delayed by an indeterminate period.
The guest places primary blame on Anthropic’s sustained doomer messaging — naming models with dramatic labels like Mythos, publicizing cybersecurity risk as a core capability concern, and years of public statements framing frontier AI as too dangerous for unrestricted release. The argument is that this campaign, combined with Anthropic’s political positioning, created the conditions the US government needed to intervene formally in model distribution. OpenAI, despite a different strategic posture and significant political investment, is now caught in the same regulatory framework. The discussion raises the concept of a “permanent underclass” — a structural bifurcation in AI access where only governments, major labs, and a handful of enterprise customers can use frontier models, while the general public is perpetually one generation behind.
The second half of the video turns practical, walking through concrete steps individuals can take to build resilience: downloading open-source model weights from Hugging Face (including DeepSeek, GLM, and Kimi) before potential government pressure forces removals, running models locally with tools like LM Studio, Ollama, or llama.cpp, and building local NAS storage for long-term model preservation. The discussion also touches on the French government already sending dataset removal requests to Hugging Face as an early signal of regulatory pressure on open model infrastructure.
📺 Source: David Ondrej · Published June 26, 2026
🏷️ Format: Interview
