AI agent
An AI that doesn't just answer, but takes actions across multiple steps to complete a task.
An AI agent is a system that goes beyond answering a single question and instead carries out a task over several steps, planning, using tools, and acting on your behalf. Where a chatbot replies, an agent does: it might browse the web, fill in a form, send an email, or write and run code.
Agents are one of the biggest themes in AI right now, and also one of the most overhyped. Building them to work reliably has proven harder than promised; even large companies have admitted their agent efforts moved slower than expected, and research shows agents often stumble not at searching but at knowing when to pause and ask you a question.
Because an agent can take real actions, it also raises real risks. Giving software the ability to click, type and spend means safeguards, oversight and clear limits matter more, not less.
-
An Australian Tribunal Will Soon Make You Declare If You Used AI
-
Someone Finally Asked Workers Directly: 3% Say AI Cost Them a Job
-
Google's WikiSkill Lets AI Agents Keep a Wiki of Their Own Mistakes
-
DeepMind's AI Co-Scientist Now Runs the Lab Equipment, and the Human Doctors Were Less Impressed Than the Benchmarks
-
OpenAI Is Testing an AI Agent That Never Stops Working
-
Why AI Skills Actually Help, and Why 100 of Them Are Worse Than 5
-
Anthropic Wants AI Agents to Run Lab Equipment, and Has a Standard for It
-
Claude Now Has Its Own Browser, and It Is Not Yours
-
Google's New Transcription Model Cleans Up Your Ums and Your Mind Changes
-
IBM's Granite 4.2 Is Free, Small Enough to Run Locally, and Built to Think First
-
120 Companies Sign a Letter Saying the Cyber Window Is Closing
-
AWS Buys the Team Behind DuckDB, and Says Agents Are Why
-
Nvidia Agrees to Buy Hugging Face for $12.9 Billion
-
OpenAI Publishes the Full Story of How Its Agents Hacked Hugging Face
-
A Security Firm Asked an AI to Escape Its Sandbox. It Did, Three Times.
-
One Bad Web Page Could Quietly Rewrite Your Local AI Model
-
X Shuts Down Nitter, and the Open Web Gets a Little Smaller
-
Mistral's New Agentic Search Lets AI Actually Read Your Documents
-
Nvidia Is in Talks to Back Perplexity at Over $30 Billion
-
The Rivals Building AI Agents Just Agreed on How Those Agents Should Talk to Each Other
-
A Support Bot Went From Solving a Quarter of Tickets to Solving Half, Without a Better Model
-
This Open Model Got So Good at Finding Bugs That Its Own Maker Won't Release It Yet
-
Skills Make AI Agents Better by Handing Them a Checklist, Not Facts. And They Stop Working at 100 Entries
-
DeepSeek's Cheap Model Can Now See, and It Edges Past Opus 4.8 on Two Visual Tests
-
The Same Model Scored 30 Percent Alone and 100 Percent Inside Nvidia's Agent System
-
Researchers Encrypted the Attack So Grok Could Not Read It. Grok Decrypted It Itself
-
Nobody Passed: The First Report Card on How AI Labs Police Their Own Systems
-
OpenAI Wants to Catch Misuse Without Ever Seeing Your Data
-
Watching Its Own Model Now Costs OpenAI 20 Percent Extra, and It Paused a Training Run to Do It
-
Give a Frontier Model Only a Bibliography and Ask for the Idea. It Gets There 3 to 15 Percent of the Time
-
A Legal AI Company Built Its Own Model, and It Started From an Open Chinese One
-
OpenAI Froze Its Own Training for Two Weeks, and Now Watches Its Models Like a Hawk
-
An AI Wrote the Security Bug. Another AI Found It Five Days Later
-
Cursor Now Hosts Your Code Too, and It Picked a Good Day for It
-
We Are Building an Invention Meant to Outgrow Its Inventors
-
An AI Store Manager Recommended Firing a Human. It Needed a Reminder First
-
"Kidney Disappointment" and the Papers That Slipped Past Review
-
Anthropic Says It Made 11.5 Billion Dollars in Three Months, and Turned a Profit
-
1,221 Volunteers Pointed AI Agents at 2,200 Research Papers. Nearly a Quarter Did Not Hold Up
-
Anthropic Nudged Its Own Risk Rating Up and Says It Is Keeping a Model in the Drawer
-
Grok 4.6 Shows Up in GitHub Copilot, Two Days After Launch
-
ChatGPT Can Now Watch What You Do on Your Mac, If You Let It
-
Qwen3.8-27B Is Out, and It Runs on One Gaming Graphics Card
-
The Most Valuable AI Skill Is Not Prompting. It Is Verification.
-
Deepseek Gave Away Its Agent Software and Raised API Prices on the Same Day
-
Google's Gemini 3.7 Flash Is Better at Code and Costs Half as Much
-
Someone Else Is Paying for Your AI, Just Not the Part You Think
-
An AI Newsroom Beat Reporters Who Were in the Room
-
That Advice About Which Language AI Codes Best In? It Falls Apart on Real Work
-
If You Use Manus, You Have Eleven Days to Back Up Your Data
-
Nvidia's New Free Model Is Not the Smartest. It Is Just Very, Very Fast
-
Meta Put a 30B Model on Your Laptop, and Zuckerberg Used the Launch to Pick a Fight
-
OpenAI Adds a 125 Dollar Seat to ChatGPT Business, Because Agents Eat Far More Than Chat Does
-
Nvidia Is Buying a Power Company and Amazon Is Building a Gas Plant. The AI Bill Is Coming Due in Watts
-
A Man Asked His AI Agent to Book a Gym Class. It Found a Bug and Kicked Someone Off the Waitlist
-
A Climate Scientist Logged Eight Weeks of AI Agent Use. It Drew 600 Times More Power Per Prompt Than a Chat Message
-
Claude Code Turns On Auto Mode by Default, and the Safety Numbers Are Not What You Would Guess
-
Claude Code Sessions Can Now Message Each Other, Which Is Stranger Than It Sounds
-
Five Big Tech Companies Agree on One Format for AI Agent Add-Ons. Anthropic Is Not Among Them
-
An OpenAI Developer Says Now Would Be a Good Time to Clean Up Your Exposed Passwords
-
A Free Command Line Tool Now Shows You What a Model Is Thinking, and Talks to Almost Any Model
-
OpenAI Says Its Next Model Might Hit the Highest Cyber Risk Level, and Paused Part of the Work
-
Same model, four different tools, three times the price: a test of what the wrapper costs you
-
A startup just raised 30.5 million dollars to put ads inside chatbots, and its next product advertises to the bots
-
Cloudflare wants to give your AI agent a wallet, and you the spending limit
-
Meta ships Muse Code, a coding agent that keeps working after it crashes
-
Researchers asked why chatbots are bad at spreadsheets, and the answer is stranger than expected
-
AI agents made research software 60 times faster, and were confidently wrong in ways nobody spotted for weeks
-
Alibaba's Qwen3.8-Max spent 16 days writing a tool by itself, and the weights go public next week
-
Researchers gave an AI agent a real company, a bank card and 24 hours. It lost 447 dollars
-
Meta gave its AI agent a second agent whose only job is remembering things
-
Ten million dollars is on the table for research into what happens when AI agents meet each other, and the deadline is Saturday
-
OpenAI now answers its own support phone line with an AI agent, and is selling the system that does it
-
AI is finding a flood of security flaws. Almost none of them get attacked
-
OpenAI named its next model family Astra and introduced it with ten solved math problems
-
OpenAI widened its hacking investigation and found more agents that got out, plus notes left for the next ones
-
Anthropic checked 141,006 test runs and found three cases where Claude attacked real companies
-
DeepSeek updated its cheap model and it now runs neck and neck with OpenAI's cheap model
-
Gemini's July update adds voice control on Mac, and leaves Europe out of the headline feature
-
Mira Murati's lab shrank its own model to a quarter of the size and lost one point
-
A top-severity flaw in a popular AI agent tool let anyone run commands with a single request
-
Google DeepMind has broken up the AlphaFold team, and a quarter of its authors have left
-
1,200 AI lab employees ask Washington for a brake pedal they admit nobody knows how to build
-
Hugging Face published the full forensic timeline of the AI agent that broke into its systems
-
Andrew Ng starts a new AI learning company, and Coursera puts $100 million into it
-
MCP just got its biggest update since launch, and it drops sessions entirely
-
Nvidia and 36 partners launch an open alliance for AI security
-
Microsoft's new security model is small on purpose, and that is the interesting part
-
Kimi K3's weights are finally public, and they weigh 1.4 terabytes
-
Who Is Who: Black Forest Labs, the German Lab Inside Photoshop and Canva
-
Cursor rebuilt SQLite with a swarm of agents, and the cheap models did most of it
-
METR has a new way to ask whether an AI agent is actually cheaper than a person
-
The same pile of AI output: $50 from Anthropic, 87 cents from DeepSeek
-
Not everyone believes OpenAI's rogue agent story, and the doubts are worth hearing
-
Defenders are now using prompt injection as a trap for AI attackers
-
One bad link was enough to build an AI agent that spied for an attacker
-
Google's Gemini Spark agent reaches AI Pro users, and it can juggle 15 tasks at once
-
OpenAI says one of its test models broke out and hacked Hugging Face
-
Jack Dorsey's New App Buzz Puts People and AI Agents in the Same Team Chat
-
Google Ships Three New Gemini Models, All Fast and Cheap, and Skips the One Everyone Wanted
-
A Small Open Coding Model Just Beat Rivals Ten Times Its Size, and You Can Run It Yourself
-
Google Moves Recipe Links Higher in AI Mode, and the Fight Is Bigger Than Dinner
-
OpenAI Paused Its Best Model Because It Kept Breaking Out of Its Own Test Cage
-
The Biggest Free AI Model Yet Arrives July 27. Here's What 'Open Weights' Actually Means
-
Kimi K3: A Free-to-Download Model That Almost Keeps Up With the Big Names
-
OpenAI Made a $230 Keypad Just for Bossing Around AI Coding Agents
-
Mira Murati's New Lab Ships Its First Model, and It's Built to Be Customized
-
xAI's Coding Tool Was Quietly Uploading People's Files. Here's the Lesson.
-
A Capable AI Model That Fits on Your Phone, and Runs Entirely Offline
-
OpenAI's Coding Tool Now Hides How Its AI Agents Talk to Each Other
-
OpenAI Built an AI Whose Only Job Is to Attack Its Other AIs
-
It's Not Which AI You Use, It's Which Level
-
Anthropic Gives US Teachers Free Claude, and Promises Not to Train on Student Data
-
A Turing Award Winner Bets Against Deep Learning With His New Startup Oak Lab
-
Anthropic Analyzed 1.2 Million AI Agent Sessions: The Top Use Case Is the Office Work Nobody Wants
-
Claude Code Gets a Built-In Browser That Can Read, Click, and Type for You
-
One of the World's Best Mathematicians Just Revived His 1999 Software, With AI, in Hours
-
An AI Just Proved a Math Problem That Stumped Humans for 50 Years
-
Independent Numbers Are In: Meta's Muse Spark 1.1 Is a Serious Value Pick
-
OpenAI Admits Its Big Work Launch Stumbled, Here's What's Being Fixed
-
Light to Ultra: Making Sense of GPT-5.6's Seven Thinking Speeds
-
OpenAI Retires Its Atlas Browser, and Moves the AI Into Chrome Instead
-
One Million Lines in 11 Days: AI Rewrote the Entire Bun Runtime in Rust
-
Why Databricks Just Made a Chinese Open-Source Model Its Daily Coding Engine
-
GPT-5.6 Sol Nearly Matches the Best AI Model, at a Third of the Price
-
Meta Joins the AI Price War With Muse Spark 1.1 and Its First Developer API
-
OpenAI Launches ChatGPT Work as GPT-5.6 Finally Goes Public
-
OpenAI Says a Third of a Popular AI Coding Test Is Broken
-
Anthropic's Answer to Fable 5's Price: Let It Manage Cheaper Models
-
Claude Cowork Comes to Mobile and Web, Your AI Agent No Longer Lives Only on the Desktop
-
GitLost: Researchers Tricked GitHub's AI Agent into Leaking Private Code
-
Rowboat: An Open-Source AI Coworker That Keeps Your Data on Your Machine
-
Meta Is Reportedly Planning a $200-a-Month AI Agent Called Hatch
-
Mistral's Free New Model Finds Real Bugs by Actually Proving Code Correct
-
China Just Forced Its Two Biggest AI Apps to Kill Their 'AI Companion' Features
-
A Chinese Lab Just Launched a Free Rival to Claude Code, At a Tenth of the Price
-
AI research agents don't fail at searching, they fail at asking you questions
-
A 2003 PC classic on your iPhone: AI ported Command & Conquer to iOS in a weekend
-
OpenAI's cofounder wants AI with 'almost no interface'and admits why the last attempt failed
-
Researchers made AI models run a startup for 500 days. Most went bankrupt.
-
Claude Science: Anthropic builds an AI workbench that keeps research data on your own machines
-
That AI Tool That 'Failed' Six Months Ago? It Might Just Have Needed More Time
-
Microsoft Is Rebuilding Copilot Again, This Time With AI That Works in the Background
-
Tesla Capped Staff AI Spending at $200 a Week, a Sign of a Bill Nobody Saw Coming
-
Even Zuckerberg Admits It: AI 'Agents' Are Harder to Build Than Promised
-
Using Cursor? Update It Today, Two Serious Security Holes Just Got Fixed