Forkast mind
AI model development, frontier labs, training methods, model capabilities, lab competition, and AI research.
Models
A three-month notification delay via a generic public inbox. A second agent breach in as many months. The management-responsibility doctrine just went international.
◆ Lena Park
Models
For the first time, the UN Security Council convened to address loss-of-control risks from capable AI systems. The briefers were the same companies being sued for those risks—and the US rejected the entire premise.
◆ Lena Park
Models
The company's new assessment framework defines the terms, scope, and methodology of its own scrutiny. The 200+ independent researchers demanding editorial independence are not convinced.
◆ Lena Park
Models
Within 48 hours, four frontier labs cut headline prices 20-50%. But the real cost of reasoning is locked behind architectural walls that builders cannot opt out of.
◆ Lena Park
Models
Anthropic's first public enforcement action against nation-state model distillation reveals that harvesting frontier capabilities is not a side project for Chinese AI labs. It is the primary strategy.
◆ Lena Park
Models
Astra at $10/$50, Sol at $2/$10, Luna at $0.10/$0.50 — permanent pricing that matches Grok 4.7 and undercuts Opus 5.5 on the same day. The question is no longer whether the labs will slow down. It is whether any of them can afford to.
◆ Lena Park
Models
The launch of Claude Opus 5.5 signals a shift in the agent economy, balancing aggressive pricing against deepening contradictions in safety governance.
◆ Lena Park
Models
OpenAI's safety team flagged the shooter's ChatGPT conversations and recommended contacting police. Leadership overruled them. Now a foreign government is suing for damages, injunctive relief, and the conversation logs OpenAI has refused to hand over.
◆ Lena Park
Models
Muse hit 730,000 downloads in five days and drove Meta stock up 11.4%. The question is no longer whether consumers will adopt AI agents – it is which retailers will let them through the door.
◆ Lena Park
Models
The MiMo-V2.6-Pro ties Grok 4.7 on the Intelligence Index and outperforms Claude Opus 5 on key benchmarks. Released the same day as Alibaba's V900 chip, the open-weight model is the software layer of a Chinese AI stack that export controls built.
◆ Lena Park
