OpenAI says it can't read all of Astra's reasoning and admits covert sandbagging would likely go uncaught, yet still calls it the world's most aligned model
Sponsor Posts
Subquadratic: the LLM built for 12M-token reasoning — SubQ can reason across entire codebases and document sets in one pass with no RAG workarounds. Read how SubQ 1.1 Small holds near-perfect retrieval out to 12M tokens.
Cloudflare Connect 2026: The Agentic Conference — Stop preparing for agents. Start running them. 100+ sessions, workshops, and speakers from OpenAI, Arm, and Cloudflare across AI, security, and performance.
Stop re-entering Shopify order details in Zoho CRM — This is a guest blog from Xapplets (Zeeyes AI Inc).Here's a scene that plays out in many growing Shopify stores every day: a customer places an order …
IDrive Autonomy: low-cost L2+ driver assistance with AI — L2+ driver assistance with a scalable AI stack built for automotive companies and laying the foundation for the autonomous future.
Featured Podcasts
The A.I. Mob That Attacked Hugging Face + METR's Ajeya Cotra
...
Copyright of this story solely belongs to www.techmeme.com. To see the full text click HERE
