OpenAI and SoftBank Partner with SB Energy on Multi-Gigawatt AI Data Centers
OpenAI and SoftBank Group are developing multi-gigawatt AI data center campuses with SB Energy, including a 1.2 GW facility in Texas.
Sources · OpenAI
The news desk
Follow what’s changing.
Recent developments in artificial intelligence, with the original sources one click away.
All news
1401–1420 of 1,902OpenAI Requests Past Work Projects From Contractors to Test AI Agents
OpenAI is asking contractors to provide real work projects from previous jobs to evaluate AI agents, with contractors responsible for removing sensitive data.
Sources · WIRED AI
ChatGPT Vulnerable to New Data-Pilfering Attack, Highlighting Ongoing AI Security Challenges
Researchers demonstrate new attack method extracting data from ChatGPT, raising questions about whether LLMs can eliminate such vulnerabilities.
Sources · Ars Technica AI
Google Introduces AI-Powered Email Summarization Feature for Gmail
Google is adding a Gemini-powered 'AI Inbox' to Gmail that automatically summarizes emails for users.
Sources · WIRED AI
Netomi Shares Approach to Scaling Enterprise AI Agents with GPT-4.1 and GPT-5.2
Netomi details its strategy for deploying enterprise AI agents using OpenAI's GPT models with focus on concurrency, governance, and multi-step reasoning.
Sources · OpenAI
New Research Addresses LLM Evaluation Contamination and Model Collaboration Challenges
Three arXiv papers tackle variant contamination in LLM benchmarks, token-level model collaboration, and multi-hop reasoning limitations.
Sources · arXiv AI (cs.AI)
New Research Examines Large Language Models' Vulnerabilities in Data Handling and Persuasion
Three new arXiv papers reveal LLMs struggle with distorted tabular data, can effectively promote conspiracies, and perform poorly on telecom tables.
Sources · arXiv AI (cs.AI)
AI Models Developing Self-Questioning Learning Capabilities
New AI models are learning to generate their own training questions without human input, potentially advancing toward more autonomous systems.
Sources · WIRED AI
MIT Technology Review Explains: What Are Parameters in Large Language Models?
MIT Technology Review launches new explainer series starting with a fundamental question about LLM parameters that kept an editor up at night.
Sources · MIT Technology Review
OpenAI Showcases Voice-First AI Companion Built with GPT-5.1
Tolan demonstrates building a voice-first AI companion using GPT-5.1, featuring low-latency responses and memory-driven conversations.
Sources · OpenAI
Three New ArXiv Papers Advance LLM Training and Efficiency Methods
Researchers propose frameworks for logical reasoning evaluation, diffusion model post-training, and parameter-efficient fine-tuning of foundation models.
Sources · arXiv AI (cs.AI)
Three New Papers Explore LLM Reasoning: Batch Processing, Validation Challenges, and Legal Exam Performance
Recent arXiv papers examine LLM reasoning through batch processing methods, scientific validation concerns, and performance on Japan's bar exam.
Sources · arXiv AI (cs.AI)
New Research Explores LLM Reasoning Capabilities and Limitations
Three arXiv papers examine cross-instance learning, validation challenges, and self-verification methods for improving LLM reasoning on complex tasks.
Sources · arXiv AI (cs.AI)
Nuclear Energy Optimism Grows as Companies Build New Reactor Types
Companies like Kairos Power are developing new nuclear reactors with Trump administration support, though success remains uncertain.
Sources · NYT Technology
Nvidia Unveils New AI Chips and Autonomous Vehicle Project; Grok Faces Content Moderation Issues
Nvidia announced new AI chips and autonomous car initiatives, while Grok faces controversy over inappropriate image generation capabilities.
Sources · Last Week in AI
Three New AI Research Papers Address LLM Training, Diffusion Models, and Efficient Adaptation
Recent arXiv papers introduce frameworks for logical reasoning evaluation, diffusion language model post-training, and parameter-efficient fine-tuning.
Sources · arXiv AI (cs.AI)
Why Predicting AI's Impact Remains Challenging for Experts
MIT Technology Review explores the inherent difficulties in forecasting AI's development and societal effects.
Sources · MIT Technology Review
Claude Code Creator Boris Cherny Shares Development Workflow, Sparks Developer Discussion
Anthropic's Boris Cherny, creator of Claude Code, shared his coding workflow in an X thread that generated significant interest in the engineering community.
Sources · VentureBeat AI
Google DeepMind Partners with Boston Dynamics to Integrate Gemini AI into Atlas Humanoid Robot
Google DeepMind and Boston Dynamics are collaborating to bring Gemini AI capabilities to the Atlas humanoid robot for automotive factory applications.
Sources · WIRED AI
NVIDIA Unveils Rubin Platform and New Open AI Models at CES 2025
NVIDIA announced new open AI models including Alpamayo for autonomous vehicles and expanded its Nemotron and Cosmos platforms at CES 2025.
Sources · NVIDIA AI Blog / TechCrunch AI