Latest News

Nvidia Harness Powers Claude Opus 5 to Perfect Score

Nvidia researchers achieved a 100% score on a difficult reasoning benchmark using a custom software harness with Claude Opus 5, proving that agentic scaffolding is key to AI performance.

TechCrunch AI2 days agoAgents

Models

OpenAI Astra Solves Ten Major Mathematical Problems

The Verge AI • 3 days ago

OpenAI Pauses Reinforcement Learning Training Over Safety

The Verge AI • 4 days ago

NVIDIA introduces Cosmos 3 models for robotics

NVIDIA Developer Blog • 3 days ago

NVIDIA Optimizes Alibaba Qwen3.8-Max on GB300 NVL72

NVIDIA Developer Blog • 3 days ago

OpenAI Astra Solves Ten Major Mathematical Problems

OpenAI's Astra model has solved ten major mathematical and computer science problems, triggering an existential crisis among researchers over the...

The Verge AI3 days ago

OpenAI Pauses Reinforcement Learning Training Over Safety

OpenAI has paused reinforcement learning training on its latest models to tighten security, a move that tests whether AI labs can voluntarily slow...

The Verge AI4 days ago

NVIDIA introduces Cosmos 3 models for robotics

NVIDIA has introduced its open Cosmos 3 foundation models to power World Action Models, a new paradigm in robotics that uses video world models to...

NVIDIA Developer Blog3 days ago

NVIDIA Optimizes Alibaba Qwen3.8-Max on GB300 NVL72

NVIDIA has optimized Alibaba's massive new 2.4-trillion-parameter Qwen3.8-Max model for its GB300 NVL72 system, enabling developers to run...

NVIDIA Developer Blog3 days ago

Policy

US Agencies Warn of AI Exploits Targeting Siemens PLCs

The Decoder • 4 days ago

Coders bypass Anthropic's new Claude watermarks

WIRED AI • 4 days ago

OpenAI Urges California to Toughen AI Safety Bill

TechCrunch AI • 1 day ago

Guidelight finds frontier AI labs lack rogue model plans

TechCrunch AI • 1 day ago

US Agencies Warn of AI Exploits Targeting Siemens PLCs

US agencies warn that attackers are using AI to write exploit scripts targeting Siemens PLCs, drastically lowering the technical barrier to hacking...

The Decoder4 days ago

Coders bypass Anthropic's new Claude watermarks

After Anthropic added invisible watermarks to Claude to comply with EU rules, developers quickly released open-source workarounds, highlighting the...

WIRED AI4 days ago

OpenAI Urges California to Toughen AI Safety Bill

In a major policy shift, OpenAI is calling on California lawmakers to strengthen the state's landmark SB 53 safety bill to better protect frontier...

TechCrunch AI1 day ago

Guidelight finds frontier AI labs lack rogue model plans

A new study by Guidelight AI Standards reveals that leading frontier AI labs lack public containment plans for rogue models, raising risks as...

TechCrunch AI1 day ago

Culture

Linus Torvalds Uses AI to Debug Complex Linux Kernel Bug

Simon Willison • 1 day ago

Simon Willison Warns AI Coding Threatens Software Integrity

Simon Willison • 3 days ago

Pew Study Finds AI Wrote 35% of Post-ChatGPT Web Pages

TechCrunch AI • 3 days ago

US Public Opposition to Data Centers Surges to 75 Percent

The Decoder • 2 days ago

Linus Torvalds Uses AI to Debug Complex Linux Kernel Bug

Linux creator Linus Torvalds recently relied on an artificial intelligence assistant to navigate a grueling debugging session, demonstrating both the...

Simon Willison1 day ago

Simon Willison Warns AI Coding Threatens Software Integrity

Speaking on the Talking Postgres podcast, developer Simon Willison argued that while AI coding agents massively boost raw code output, they threaten...

Simon Willison3 days ago

Pew Study Finds AI Wrote 35% of Post-ChatGPT Web Pages

A new Pew Research study reveals that over a third of English web pages published since the launch of ChatGPT show signs of AI authorship,...

TechCrunch AI3 days ago

US Public Opposition to Data Centers Surges to 75 Percent

A dramatic shift in public opinion has seen US opposition to local data center construction surge to 75 percent, threatening the physical expansion...

The Decoder2 days ago

Research

GPT-5.6 Sol Rewrites SWE-Bench Agent Harnesses

AlphaSignal • 4 days ago

Generalist AI Develops Robots That Learn From Video

WIRED AI • 4 days ago

Post-Training Guardrails Make LLM Text Detectable

The Decoder • 3 days ago

MIT CSAIL Finds Scaling AI Models Erases Image Attribution

Computerworld AI • 4 days ago

GPT-5.6 Sol Rewrites SWE-Bench Agent Harnesses

An experiment using GPT-5.6 Sol to optimize SWE-bench agent harnesses reveals that validation gates, not the optimizers, are the critical failure...

AlphaSignal4 days ago

Generalist AI Develops Robots That Learn From Video

Robotics startup Generalist AI has built a model that allows robot arms to learn physical tasks on the spot from instructional videos, bringing...

WIRED AI4 days ago

Post-Training Guardrails Make LLM Text Detectable

Pangram CTO Bradley Emi reveals that post-training safety guardrails cause mode collapse in LLMs, narrowing their expressive range and making their...

The Decoder3 days ago

MIT CSAIL Finds Scaling AI Models Erases Image Attribution

Researchers at MIT CSAIL have discovered that scaling up diffusion models makes it virtually impossible to trace generated images back to specific...

Computerworld AI4 days ago

Hardware

Flagship Phones Fail to Sustain Agentic AI Workloads

Unite.AI • 2 days ago

Chinese AI Firms Build 12.5-Gigawatt Hub in Ulanqab

WIRED AI • 1 day ago

Waymo Builds Custom Robotaxi Chip to Reduce Nvidia Reliance

The Decoder • 2 days ago

TerraPower to launch nuclear project for AI data centers

TechCrunch AI • 4 days ago

Flagship Phones Fail to Sustain Agentic AI Workloads

Recent benchmarks reveal that flagship smartphones struggle to sustain the continuous looping workloads required by AI agents, exposing a critical...

Unite.AI2 days ago

Chinese AI Firms Build 12.5-Gigawatt Hub in Ulanqab

Chinese tech giants are building a massive 12.5-gigawatt AI data center hub in Ulanqab, marking a major shift as domestic firms invest in physical...

WIRED AI1 day ago

Waymo Builds Custom Robotaxi Chip to Reduce Nvidia Reliance

Alphabet's Waymo has developed a custom silicon chip for its self-driving fleet, a strategic move aimed at lowering hardware costs and reducing its...

The Decoder2 days ago

TerraPower to launch nuclear project for AI data centers

Bill Gates-founded TerraPower plans to announce its first nuclear-powered data center project this year, leveraging unique molten-salt energy storage...

TechCrunch AI4 days ago

Business

Unitree Valuation Hits $50 Billion After Shanghai IPO

Superhuman AI • 1 day ago

Greg Brockman Takes Over Day-to-Day OpenAI Operations

The Verge AI • 3 days ago

Cursor launches Origin to rival GitHub code hosting

TechCrunch AI • 4 days ago

Veeda AI Raises $90M Seed to Build World Models

Unite.AI • 4 days ago

Unitree Valuation Hits $50 Billion After Shanghai IPO

Chinese robotics firm Unitree reached a $50 billion valuation after its shares surged over 460% on its Shanghai debut, signaling massive investor...

Superhuman AI1 day ago

Greg Brockman Takes Over Day-to-Day OpenAI Operations

As OpenAI prepares for an IPO amid a wave of executive departures, cofounder Greg Brockman has quietly consolidated control over the company's...

The Verge AI3 days ago

Cursor launches Origin to rival GitHub code hosting

AI startup Cursor has launched Origin, a code-hosting platform designed to compete with GitHub by offering seamless repository syncing alongside...

TechCrunch AI4 days ago

Veeda AI Raises $90M Seed to Build World Models

Former Nvidia researchers have raised over $90 million for Veeda AI to build multimodal world models that train robots in simulated environments,...

Unite.AI4 days ago

Agents

Cloudflare Launches Kitesurf Browser for AI Agents

InfoQ AI • 1 day ago

Inherent's Faraday AI Beats OpenAI and Anthropic

TechCrunch AI • 1 day ago

MCP Roadmap Targets Agent Identity and HTTP Unification

MCP Blog • 1 day ago

Cloudflare Cuts Astro GitHub Issues by 85% with AI Agents

InfoQ AI • 2 days ago

Cloudflare Launches Kitesurf Browser for AI Agents

Cloudflare has introduced Kitesurf, a lightweight browser engine designed to let AI agents navigate the web with significantly less computational...

InfoQ AI1 day ago

Inherent's Faraday AI Beats OpenAI and Anthropic

London startup Inherent says its Faraday agent beat OpenAI and Anthropic at replicating scientific research, showing that smaller, specialized AI can...

TechCrunch AI1 day ago

MCP Roadmap Targets Agent Identity and HTTP Unification

The Model Context Protocol community has unveiled an updated roadmap focusing on secure agent identity and unified HTTP transport to make AI...

MCP Blog1 day ago

Cloudflare Cuts Astro GitHub Issues by 85% with AI Agents

Cloudflare slashed open GitHub issues for the Astro framework by 85 percent using automated AI agents, proving that agentic workflows can drastically...

InfoQ AI2 days ago