This Stratechery weekly digest highlights Anthropic's new model, claimed to be the most powerful in the world but deemed too dangerous for public release. The roundup also covers OpenAI's acquisition of the TBPN technology podcast and analysis of the New York Times' successful digital strategy built on early paywall adoption and human expertise differentiation.
Safety
2026.15: Myth and Mythos
Anthropic released its most powerful model yet but restricted public access, citing safety concerns that outpace deployment readiness — underscoring the widening gap between frontier model capabilities and the guardrails regulators and labs deem necessary.
Friday, April 10, 2026 12:00 PM UTC2 MIN READSOURCE: StratecheryBY sys://pipeline
Tags
safety
/// RELATED
PolicyApr 7
OpenAI #16: A History and a Proposal
New Yorker investigates Sam Altman's trustworthiness while OpenAI proposes its own regulatory framework — a credibility challenge for industry self-governance.
StrategyApr 7
OpenAI is a drama company. Will that hurt its IPO chances? And Anthropic tries to get ahead of the cyber risks its own models are accelerating
As OpenAI's governance turmoil and $600B spending disputes threaten its IPO timeline, Anthropic preemptively launches Project Glasswing—offering early access to its Mythos model to detect zero-days before AI-enabled cyberattacks proliferate.