Latest

OpenAI Shuts Down Sora, Refocuses on Code AGI and Model Spud

OpenAI **discontinues Sora and all video generation products** to redeploy compute resources toward competing with Anthropic in enterprise coding and knowledge work. CEO Sam Altman **narrows his role to focus on capital, supply chains, and data centers**; Fiji Simo's product division becomes 'AGI Deployment' team. New model 'Spud' completes pre-training with expectations to 'accelerate the economy'; **Disney cancels $1B investment partnership** following Sora shutdown.

Autism-Therapy Firm That Was Paid $340k per Patient Is Barred from Medicaid

Chonkify – compression for RAG and Agents that outperforms LLMLingua by ~4 times

Agent Identity System

The U.S. Said It Helped Bomb a Drug Camp. It Was a Dairy Farm

DarkSword Exploit Chain, Unpackaged

Rich Content AI Chat

Claude Max Guest Passes: Free Week Access to Claude Code and Cowork

Claude Max now offers **guest passes for free one-week access** to Claude Code and Cowork features. Limited availability: **only 3 guest passes being distributed** by klöss for testing premium Claude capabilities. **Opportunity to test advanced Claude features** without subscription commitment before deciding to upgrade.

70% of new software engineering papers on ArXiv are LLM related

Show HN: Book Grounded AI Learning

I constantly ask Gemini&#x2F;ChatGPT&#x2F;Claude about books on a certain topic, then ask the llm to give me a summary of the book its chapters - this helps me: 1. keep the llm grounded as a real expert (instead of saying pretend to be an engineering expert, give it a distributed systems book and then ask opinions on it) 2. helps me discover and buy books! Once i go through book concepts and chat with llm about it, i&#x27;m more convinced to buy it and dig deeper on low level on it<p>So i built an app through Replit on it, try it on<p><a href="https:&#x2F;&#x2F;teach-me.replit.app&#x2F;" rel="nofollow">https:&#x2F;&#x2F;teach-me.replit.app&#x2F;</a><p>Cheers!

AI got the blame for the Iran school bombing. The truth is more worrying

Cost-Benefit Analysis of LLM-Based Content Ranking at Scale

Exploring whether **using Grok to scan and rank posts** is cost-effective compared to traditional algorithms at scale. Questions the **financial viability of LLM-based content moderation** versus conventional algorithmic approaches for large platforms. Highlights need for **clearer system metrics to determine** which ranking method delivers better ROI for content platforms.