中文 English Français Русский Español العربية

OpenAI’s new reasoning technique alarms AI safety experts

Russell Brandom TechCrunch 2026-09-03 04:19 19 views

OpenAI’s new Astra model will use “recurrent depth,” a technique that allows the model to operate outside of the sequential thinking that characterizes most reasoning models.

Read original: https://techcrunch.com/2026/09/02/openais-new-reasoning-technique-alarms-ai-safety-experts/

My Favorites →

← Previous MapQuest is now the No. 1 US app after bucking Trump’s ‘Lake America’ renaming Next → GitOps in 2026: Why Pull Requests Are Taking Over Cloud Operations

OpenClaw Plugin: Connect WAIaaS to LangChain, CrewAI, and Any AI Framework

OpenClaw Plugin: Connect WAIaaS to LangChain, CrewAI, and Any AI Framework Your AI agent can browse the web, write code, and manage files — but can it swap tokens? The OpenClaw plugin is WAIaaS's answer to that question: a drop-in wallet toolkit that connects your existing AI agent framework to re

AI & ML 09-15 60 Dev.to (EN Zone)

Why Giving AI Agents More Context Can Make Them Worse

Most AI agent demos start with the model. We ended up spending far more time thinking about what gets put around the model. I run 23 holiday lets and have been building AI agents into the day-to-day operation. Guest messaging sounds like one of the easier problems: Guest sends a message. Give th

AI & ML 09-15 49 Dev.to (EN Zone)

How to Optimize Development Cycles with AI: Numbers, Tools & Tactics [2026]

Originally published at nlocoding.com 58% of software projects still miss deadlines, even when AI tools are in the mix. (Standish Group, 2026) Why does this matter? Because the market doesn't wait. A late launch means you lose users, revenue, and trust—fast. In 2026, the average cost of a one-m

AI & ML 09-15 51 Dev.to (EN Zone)

How to Stop Leaking API Keys and PII to LLMs: Zero-Trust In-Memory Sanitization for Cursor & Claude

If you are using AI coding assistants like Cursor, Windsurf, Claude Desktop, or ChatGPT, you have probably pasted a snippet containing sensitive data at least once: Hardcoded AWS or OpenAI API keys Database connection strings (postgres://user:password@host/db) JWT tokens and private RSA keys Real

AI & ML 09-15 49 DEV Community

Comments0

No comments yet — be the first