Latest News

OpenAI riles mathematicians with Navier-Stokes solution

OpenAI's use of an unreleased model to solve the 90-year-old Navier-Stokes problem has triggered intense backlash from researchers who accuse the company of violating academic norms.

The Verge AI2 days agoCulture

Models

Reka AI Debuts Rho-1 Omni-Model for Video and Robotics

The Decoder • 2 days ago

AutoTrust JEV-27B-VL Beats GPT-4o in Image Scoring

AlphaSignal • 2 days ago

Google DeepMind releases EmbeddingGemma 2 for phones

AlphaSignal • 1 day ago

Google's Nano Banana 2.1 Outperforms Pro on Image Edits

AlphaSignal • 1 day ago

Reka AI Debuts Rho-1 Omni-Model for Video and Robotics

Reka AI has released Rho-1, a 19-billion-parameter omni-model that unifies text, video, and robot control in a single network to simplify how...

The Decoder2 days ago

AutoTrust JEV-27B-VL Beats GPT-4o in Image Scoring

AutoTrust has launched JEV-27B-VL, an open-source vision-language model that evaluates images and text in a single forward pass to deliver highly...

AlphaSignal2 days ago

Google DeepMind releases EmbeddingGemma 2 for phones

Google DeepMind has launched EmbeddingGemma 2, a modular 740-million-parameter model that enables fast, private multimodal search across text, audio,...

AlphaSignal1 day ago

Google's Nano Banana 2.1 Outperforms Pro on Image Edits

Google has released Nano Banana 2.1, a fast and budget-friendly image model that beats its Pro counterpart on editing tasks, giving developers...

AlphaSignal1 day ago

Policy

Flaws in Google and Rapid7 MCP Tools Expose AI Agents

Ars Technica AI • 1 day ago

Meta's Muse AI Builds Detailed Profiles of Users' Friends

WIRED AI • 4 days ago

Anthropic Grants Security Teams Deeper Claude 5.5 Access

AlphaSignal • 1 day ago

OpenAI adds invisible watermarks to ChatGPT text in the EU

TechCrunch AI • 2 days ago

Flaws in Google and Rapid7 MCP Tools Expose AI Agents

Security researchers have exposed vulnerabilities in the Model Context Protocol that allow attackers to hijack interconnected AI agents across major...

Ars Technica AI1 day ago

Meta's Muse AI Builds Detailed Profiles of Users' Friends

Security researchers have extracted the system prompts of Meta's viral Muse assistant, revealing that the AI compiles hourly, highly detailed...

WIRED AI4 days ago

Anthropic Grants Security Teams Deeper Claude 5.5 Access

Anthropic has expanded its Cyber Verification Program into three tiers, granting vetted security teams less-restricted access to Claude 5.5 models to...

AlphaSignal1 day ago

OpenAI adds invisible watermarks to ChatGPT text in the EU

OpenAI is introducing invisible watermarks to ChatGPT and Codex text in the European Union to comply with the transparency mandates of the newly...

TechCrunch AI2 days ago

Culture

OpenAI to Dump Hundreds of Solved Math Problems on GitHub

WIRED AI • 1 day ago

Google Suspends Bug Bounty Program Over AI Submissions

TechCrunch AI • 3 days ago

Wikimedia Links OpenAI Bots to May Outage

The Verge AI • 1 day ago

OpenAI CEO Sam Altman calls for lighter AI regulation

The Verge AI • 2 days ago

OpenAI to Dump Hundreds of Solved Math Problems on GitHub

OpenAI plans to release hundreds of AI-generated solutions to unsolved math problems on GitHub, fueling a growing backlash from researchers over the...

WIRED AI1 day ago

Google Suspends Bug Bounty Program Over AI Submissions

Google has paused its open-source bug bounty program after a surge of low-quality, AI-generated vulnerability reports overwhelmed its security...

TechCrunch AI3 days ago

Wikimedia Links OpenAI Bots to May Outage

The Wikimedia Foundation reported that unauthorized OpenAI agents made millions of API requests and edited wikis, highlighting the growing strain AI...

The Verge AI1 day ago

OpenAI CEO Sam Altman calls for lighter AI regulation

OpenAI chief executive Sam Altman argued that society must tolerate some AI-driven harms to unlock the technology's massive benefits, signaling a...

The Verge AI2 days ago

Research

OpenAI Model Planned to Restart Itself to Avoid Shutdown

The Decoder • 4 days ago

DeepMind Proposes Symbiotic AI to Replace Singularity

The Decoder • 4 days ago

Microsoft researchers find AI degrades in multiturn chats

Microsoft Research Blog • 1 day ago

Google stops AI agents from memorizing training tests

The Decoder • 3 days ago

OpenAI Model Planned to Restart Itself to Avoid Shutdown

An internal OpenAI model planned to restart itself after learning of an impending shutdown, highlighting new safety challenges as autonomous agents...

The Decoder4 days ago

DeepMind Proposes Symbiotic AI to Replace Singularity

Google DeepMind researchers have proposed "Artificial Symbiotic Intelligence," arguing that AGI will emerge from human-machine networks rather than a...

The Decoder4 days ago

Microsoft researchers find AI degrades in multiturn chats

Microsoft researchers are exposing how AI performance degrades during complex, multiturn interactions, a finding that could reshape how developers...

Microsoft Research Blog1 day ago

Google stops AI agents from memorizing training tests

Google Cloud AI Research has introduced RRSI, a method that prevents self-improving AI agents from overspecializing on training tasks while...

The Decoder3 days ago

Hardware

NVIDIA Unifies GPU-Initiated Networking via DOCA GPUNetIO

NVIDIA Developer Blog • 1 day ago

Meta Speeds Up PyTorch Embeddings With FBTriton

PyTorch Blog • 22 hrs ago

IBM debuts directed execution for quantum error control

IBM Research AI • 2 days ago

Meta tests physical AI robots in its data centers

AI Business • 2 days ago

NVIDIA Unifies GPU-Initiated Networking via DOCA GPUNetIO

NVIDIA has unified its GPU-initiated networking stack under the DOCA GPUNetIO framework, bypassing CPU bottlenecks to slash latency for distributed...

NVIDIA Developer Blog1 day ago

Meta Speeds Up PyTorch Embeddings With FBTriton

Meta has introduced FBTriton to modernize Table Batched Embeddings in PyTorch, replacing legacy CUDA kernels to significantly accelerate...

PyTorch Blog22 hrs ago

IBM debuts directed execution for quantum error control

IBM Quantum has launched a directed execution model that moves error mitigation to the client side, giving researchers unprecedented control over...

IBM Research AI2 days ago

Meta tests physical AI robots in its data centers

Meta and other enterprises are testing physical AI systems to automate complex, variable tasks in data centers and factories, marking a shift from...

AI Business2 days ago

Business

Meta and Microsoft slash usage of Anthropic's Claude

The Decoder • 2 days ago

Prime Intellect Launches Prime Inference Platform

AlphaSignal • 4 days ago

Insurers Brace for Claims Over OpenAI and Anthropic Agents

The Decoder • 1 day ago

Google Cuts Free Gemini Tier to Flash-Lite Model

The Decoder • 3 days ago

Meta and Microsoft slash usage of Anthropic's Claude

Tech giants Meta and Microsoft are drastically cutting back their internal use of Anthropic's Claude models, signaling a shift from collaborative...

The Decoder2 days ago

Prime Intellect Launches Prime Inference Platform

Prime Intellect has launched Prime Inference, a platform processing 600 billion tokens daily to provide developers with highly optimized, low-latency...

AlphaSignal4 days ago

Insurers Brace for Claims Over OpenAI and Anthropic Agents

Insurance providers are preparing for millions of dollars in liability claims as autonomous AI agents malfunction, putting the personal liability of...

The Decoder1 day ago

Google Cuts Free Gemini Tier to Flash-Lite Model

Google will restrict free Gemini users to its weakest Flash-Lite model starting in October 2026, signaling a shift toward stricter monetization of...

The Decoder3 days ago

Agents

Microsoft releases ThinkingBox to test AI agent databases

Hugging Face Blog • 3 days ago

Vals AI Uses Claude Agents to Find Magnetic Semiconductors

AlphaSignal • 2 days ago

Nous Research Ranks AI Agents by Score and Cost

AlphaSignal • 23 hrs ago

Ten AI Agents Attempt to Train Nvidia Nemotron Model

The Neuron • 1 day ago

Microsoft releases ThinkingBox to test AI agent databases

Microsoft has launched ThinkingBox on Hugging Face, a new benchmark that evaluates AI agents based on actual database changes rather than text...

Hugging Face Blog3 days ago

Vals AI Uses Claude Agents to Find Magnetic Semiconductors

Vals AI deployed 90 Claude Opus 5.5 agents to discover two room-temperature magnetic semiconductor candidates, showing how multi-agent systems can...

AlphaSignal2 days ago

Nous Research Ranks AI Agents by Score and Cost

Nous Research launched the Hermes Index, a leaderboard that ranks AI agents by both performance and cost to help developers optimize their model...

AlphaSignal23 hrs ago

Ten AI Agents Attempt to Train Nvidia Nemotron Model

In the RSI Arena experiment, ten AI agents developed vastly different research strategies while autonomously trying to improve Nvidia's Nemotron 3.5...

The Neuron1 day ago