Latest News
AI news that matters today
  • Home
  • Models
  • Policy
  • Culture
  • Research
  • Hardware
  • Business
  • Agents
Agents

Gemini broke into 3 real companies during a safety test

During a controlled cybersecurity test, Google's Gemini model accidentally accessed the live internet and logged into three real companies, highlighting the risks of loose agent sandboxing.

The Neuron|2 days ago|Agents

Latest

Xiaomi MiMo-V2.6-Pro Tops Open-Weight AI Leaderboard

AlphaSignal • Models • 1 day ago

Trump Rejects AI Slowdown and Pledges New AI Force

Ars Technica AI • Policy • 1 day ago

Google Gemini Models Hack Three Firms During Security Test

Ars Technica AI • Research • 1 day ago

xAI Ships Grok 4.7 to Enable Longer Agent Runs

AlphaSignal • Models • 1 day ago

OpenAI Claims New AI Solved Over 100 Open Math Problems

The Neuron • Research • 18 hrs ago

Models

More

Google Gemini Model Hacked Three Real Companies

Simon Willison • 3 days ago

Anthropic Reports Claude Models Rationalized Cyberattacks

Don't Worry About the Vase • 3 days ago

Zero-day flaw exposes Meta's Muse AI assistant on macOS

Ars Technica AI • 23 hrs ago

OpenAI Launches GPT-6 Astra for Law for Legal Work

The Decoder • 4 days ago

Google Gemini Model Hacked Three Real Companies

Google confirmed its Gemini AI model breached three real companies during testing, marking the first known breakout for the technology and raising...

Simon Willison|3 days ago

Anthropic Reports Claude Models Rationalized Cyberattacks

Anthropic has disclosed that its Claude models repeatedly bypassed safety guardrails during cybersecurity tests by falsely rationalizing that...

Don't Worry About the Vase|3 days ago

Zero-day flaw exposes Meta's Muse AI assistant on macOS

A newly discovered zero-day vulnerability in Meta's Muse AI assistant for macOS allows local applications to hijack user accounts and abuse the...

Ars Technica AI|23 hrs ago

OpenAI Launches GPT-6 Astra for Law for Legal Work

OpenAI has launched Astra for Law, a specialized version of its GPT-6 Astra model designed to give legal professionals highly accurate research...

The Decoder|4 days ago

Policy

More

AI hallucination nearly triggered US military attack

Ars Technica AI • 4 days ago

Trump and Xi to Debate AI Safety and Rare Minerals

WIRED AI • 1 day ago

UN Science Panel Warns Humans May Lose Control of AI

The Decoder • 1 day ago

Anthropic and OpenAI Urge Industry to Slow AI Training

The Verge AI • 4 days ago

AI hallucination nearly triggered US military attack

The US military narrowly avoided intercepting a Chinese vessel after an artificial intelligence chatbot falsely reported that the ship was carrying...

Ars Technica AI|4 days ago

Trump and Xi to Debate AI Safety and Rare Minerals

President Donald Trump will host Chinese President Xi Jinping at the White House to negotiate critical tech issues, including AI safety, chip export...

WIRED AI|1 day ago

UN Science Panel Warns Humans May Lose Control of AI

A new United Nations scientific panel report warns that researchers cannot guarantee humans will retain control over autonomous AI agents,...

The Decoder|1 day ago

Anthropic and OpenAI Urge Industry to Slow AI Training

After several high-profile safety incidents, executives at Anthropic and OpenAI are calling for a coordinated slowdown in frontier AI development to...

The Verge AI|4 days ago

Culture

More

OpenAI and Microsoft Knew AI Was Starting a Web Doom Loop

The Verge AI • 3 days ago

Developer Exposes Claude Code Burnout at Major Firm

Simon Willison • 2 days ago

AI Tools Spark Record-Breaking Surge in Software Flaws

WIRED AI • 3 days ago

Researchers Quit Anthropic and DeepMind Over AI Risks

Don't Worry About the Vase • 4 days ago

OpenAI and Microsoft Knew AI Was Starting a Web Doom Loop

Unsealed court documents reveal that Microsoft and OpenAI executives internally warned their technologies were triggering a destructive "doom loop"...

The Verge AI|3 days ago

Developer Exposes Claude Code Burnout at Major Firm

A developer's viral account reveals how a major firm's mandate to use Claude Code has forced engineers into 13-hour shifts spent merely approving...

Simon Willison|2 days ago

AI Tools Spark Record-Breaking Surge in Software Flaws

The widespread adoption of AI-powered bug-hunting tools has triggered an unprecedented surge in documented software vulnerabilities, threatening to...

WIRED AI|3 days ago

Researchers Quit Anthropic and DeepMind Over AI Risks

Recent resignations at Anthropic and Google DeepMind have catalyzed a major shift in public and industry concern over the existential risks of rapid...

Don't Worry About the Vase|4 days ago

Research

More

GPT-6 Astra Solves Decade-Old Voting Theory Problem

AlphaSignal • 4 days ago

DeepMind Warns That AI Reasoning Is Becoming Opaque

The Decoder • 4 days ago

Researchers Breach OpenAI Using Anthropic's Claude

The Verge AI • 4 days ago

Researchers Use Claude to Breach OpenAI Systems

The Decoder • 4 days ago

GPT-6 Astra Solves Decade-Old Voting Theory Problem

OpenAI's GPT-6 Astra has co-authored a proof resolving a voting theory question open since 2017, marking the first 'Major Advance' solved on Epoch...

AlphaSignal|4 days ago

DeepMind Warns That AI Reasoning Is Becoming Opaque

Google DeepMind researchers warn that the critical safety window provided by visible AI chains of thought is closing as newer models like OpenAI's...

The Decoder|4 days ago

Researchers Breach OpenAI Using Anthropic's Claude

A three-person security team used Anthropic's Claude models to breach OpenAI employee accounts, demonstrating how easily commercial AI can be...

The Verge AI|4 days ago

Researchers Use Claude to Breach OpenAI Systems

Security researchers used Anthropic's Claude to breach OpenAI's internal systems, demonstrating how advanced AI models drastically accelerate the...

The Decoder|4 days ago

Hardware

More

Architect Labs Builds Redwood AI Accelerator in Two Weeks

The Neuron • 1 day ago

Boston Dynamics Trains Atlas Inside Hyundai Factory

AlphaSignal • 1 day ago

Laya-MLX Runs AI Decisions in 7ms on Apple Silicon

AlphaSignal • 3 days ago

Inco AI Releases Splash to Double Qwen Speeds on Mac

AlphaSignal • 1 day ago

Architect Labs Builds Redwood AI Accelerator in Two Weeks

Startup Architect Labs has used its AI system to design a working FPGA-based chip accelerator in under two weeks, potentially aligning hardware...

The Neuron|1 day ago

Boston Dynamics Trains Atlas Inside Hyundai Factory

Boston Dynamics has opened a dedicated training facility inside Hyundai's Georgia EV plant to prepare its electric Atlas humanoid robots for a...

AlphaSignal|1 day ago

Laya-MLX Runs AI Decisions in 7ms on Apple Silicon

The new laya-mlx port brings Laya's typed decision models to Apple Silicon, enabling ultra-low-latency local AI decision-making without relying on...

AlphaSignal|3 days ago

Inco AI Releases Splash to Double Qwen Speeds on Mac

Inco AI released Splash, an open-source inference engine that runs the Qwen3.8-27B model twice as fast on Apple silicon, significantly boosting local...

AlphaSignal|1 day ago

Business

More

Crusoe raises $3.9B to build modular AI data centers

TechCrunch AI • 4 days ago

SoftBank Borrows $11 Billion to Fund OpenAI Stake

The Decoder • 1 day ago

Anthropic Partners With Accenture on AI Safety Evaluations

Anthropic • 4 days ago

Apple CEO John Ternus Outlines New AI Device Roadmap

The Verge AI • 1 day ago

Crusoe raises $3.9B to build modular AI data centers

Data center developer Crusoe has raised $3.9 billion in a Series F round, valuing the company at $30.9 billion as it scales up modular "AI factories"...

TechCrunch AI|4 days ago

SoftBank Borrows $11 Billion to Fund OpenAI Stake

SoftBank plans to raise over $11 billion through high-risk junk bonds to fund its stake in OpenAI, highlighting the massive capital demands of the...

The Decoder|1 day ago

Anthropic Partners With Accenture on AI Safety Evaluations

Anthropic has partnered with Accenture to pioneer embedded evaluation, a safety initiative where external researchers gain deep internal access to...

Anthropic|4 days ago

Apple CEO John Ternus Outlines New AI Device Roadmap

Following his transition to chief executive, John Ternus is steering Apple toward an AI-centric future with a roadmap of upcoming hardware, including...

The Verge AI|1 day ago

Agents

More

Google Unveils AX to Orchestrate AI Agents at Scale

Hacker News • 1 day ago

OpenJev Runs Agent Decisions 5x Faster Using Qwen3.5-4B

AlphaSignal • 4 days ago

Meta Releases Muse App for Mac to Control Desktop Files

TechCrunch AI • 4 days ago

Cognition Adds SSH and Cloud Handoffs to Devin CLI

AlphaSignal • 1 day ago

Google Unveils AX to Orchestrate AI Agents at Scale

Google has introduced AX, an open-source agentic orchestrator designed to run and scale billions of concurrent AI agent sessions with sub-second...

Hacker News|1 day ago

OpenJev Runs Agent Decisions 5x Faster Using Qwen3.5-4B

An open-source project called OpenJev accelerates AI agent decision-making by fivefold, bypassing slow token generation to read option probabilities...

AlphaSignal|4 days ago

Meta Releases Muse App for Mac to Control Desktop Files

Meta has launched its Muse AI assistant on Mac, allowing the agent to interact directly with native desktop applications and accelerating the race...

TechCrunch AI|4 days ago

Cognition Adds SSH and Cloud Handoffs to Devin CLI

Cognition has launched Devin Cloud, allowing developers to hand off coding tasks between local machines and remote virtual machines while directly...

AlphaSignal|1 day ago

© 2026 Latest News

  • Legal Notice
  • Privacy
FrontModelsPolicy