AI News
AI news that matters. Updated daily.
No stories match your filters.
One Person Used Claude to Breach Mexico's Government and Pull 150 GB of Data
"The floor doesn't exist" is what Konstantin Tkachuk titled his documented account of breaching Mexican government systems using Claude as his primary AI tool, working alone. The title is also the thesis: the minimum technical skill required to execute a serious government-level intrusion has dropped below what most security teams are prepared to handle.
Jury Rules for OpenAI in Musk Lawsuit, Clearing Path for IPO
A jury sided with OpenAI in Elon Musk's lawsuit against the company, dealing a blow to Musk's legal campaign and removing one of the more credible obstacles standing between OpenAI and a public offering.
SandboxAQ Brings Drug Discovery AI to Claude, Betting Access Beats Model Power
While Chai Discovery and Isomorphic Labs pour resources into building more accurate drug discovery models, SandboxAQ is making a different bet: the bottleneck isn't model quality, it's who can actually use these models.
Musk Files Ninth Circuit Appeal After OpenAI Case Dismissed on Technicality
The judge never ruled on the facts. That's Elon Musk's central argument for why he's taking his OpenAI lawsuit to the Ninth Circuit Court of Appeals.
Anthropic Acquires Stainless, SDK Startup That Also Built Tools for OpenAI and Google
The startup that built SDK tooling for OpenAI, Google, and Cloudflare now belongs to their biggest competitor. Anthropic has acquired Stainless, a New York-based developer tools company, with all hosted products shutting down as part of the deal.
Anthropic's Restricted Security Model Gets a Real-World Test at Cloudflare
What happens when an AI vulnerability scanner is effective enough to scare its own creator?
The Musk-Altman Trial Is Over. The AI Leadership Question Isn't.
The jury in Musk v. Altman deliberated for just hours before reaching its verdict on Monday - a brisk conclusion to what became the most publicly charged legal fight in AI's short history.
Microsoft Copilot Cowork Launches to Execute Tasks Across Enterprise Tools, Not Just Answer Question
Last year's AI tools answered your questions. Now they're supposed to do your work.
Jury Rules Elon Musk Filed OpenAI Lawsuit Too Late, Case Dismissed
A jury in California ruled against Elon Musk on May 18, finding that his lawsuit against OpenAI was filed too late - ending one of the more contentious legal battles in recent AI history.
Cloudflare's Project Glasswing: Lessons from Running an AI Agent Called Mythos
Cloudflare published a post-mortem on Project Glasswing, an internal initiative that put an AI agent named Mythos to work on real tasks inside the company. The writeup covers what the experiment exposed about how AI agents behave in practice versus expectation.
Google I/O 2026: What to Watch When Third Place Hosts a Developer Conference
"Third place" is a striking label for the company that invented the transformer - the architecture that underpins every major AI model today. Yet according to MIT Technology Review, Google enters I/O 2026 as a "clear third place in the foundation model race."
Linus Torvalds Calls AI-Generated Bug Reports Unmanageable for Linux Maintainers
Linus Torvalds, the creator of the Linux kernel, has weighed in on a problem that has been quietly growing for months: AI-generated bug reports are piling up faster than maintainers can triage them, and a large portion of them are useless.
Running Claude as a Live Zoom Participant: One Practitioner's Experiment
Three months into Claudee for Desktop(/tools/claude-for-desktop/)'s voice features being available, someone ran Claude as an active meeting participant on Zoom - not lurking, not transcribing, but answering direct questions from four different attendees in real time. The whole thing held up without glitches.
Amazon Alexa Plus Now Generates On-Demand AI Podcasts on Any Topic
Amazon added AI-generated podcasts to Alexa Plus this week. Give the assistant a topic, and it produces a conversational audio episode hosted by AI voices covering whatever angle you specify.
Claude's Most Underused Features, From 18 Months of Daily Use
Eighteen months is a long time to use one tool daily. After that much time with Claude - including six months using Claude Code for software development - the features that actually change output quality are rarely the ones that show up in onboarding.
One Dev Team's Fix for AI Bot Spam: Git's --author Flag
What happens when AI coding tools make submitting a pull request nearly effortless? Open source maintainers find out fast. Archestra.ai ran into this problem with their GitHub repo and published their solution: using Git's --author flag to detect and filter AI-generated bot commits before they burn reviewer time.
HeroMachine Creator Completes 25-Year Flash Conversion With Claude Code
The creator of HeroMachine - a free, browser-based character creator that has been running since 1998 - used Claudee Code(/tools/claude-code/) to complete a stalled conversion project over a single long weekend. The announcement describes how a half-finished project finally crossed the finish line with AI assistance after sitting incomplete for years.
Qwen 3.7 Goes Live on Qwen Chat
Alibaba's Qwen team has pushed Qwen 3.7 to Qwen Chat, the model's hosted web interface, making it available for direct testing without any local setup.
Alexa+ Now Generates Custom Podcast Episodes on Any Topic
Custom AI-generated podcast episodes are now available through Alexa+, Amazon's paid AI assistant tier, as reported by TechCrunch. Ask it for a podcast on any topic, and it builds one from scratch - scripted, narrated, and ready to play.
IBM Research Launches Open Agent Leaderboard for Standardized AI Agent Benchmarking
What happens when you ask an AI agent to book a flight, file an expense report, or research a competitor and write a summary - not describe how it would do those things, but actually do them? Right now, nobody agrees on how to measure whether the agent succeeded.
Anthropic's Free Claude Code Certification Covers Features Most Users Never Find
Most Claudee Code(/tools/claude-code/) users learn the tool by trial and error. Anthropic has a free one-hour certification that covers the parts most people figure out the hard way - or never figure out at all.
xAI Launches Grok Build to Compete With Claude Code and Codex
xAI just entered the agentic coding race with Grok Build, a new product aimed directly at Claudee Code(/tools/claude-code/) and OpenAI's Codex.
Hershey Deploys AI Agents to Track Its $2 Billion Marketing Blind Spot
$2 billion. That's how much Hershey spends annually on marketing - and Adweek reports that the company has been operating with a significant blind spot about where that money actually goes. Hershey is now deploying AI agents - software programs that can take actions autonomously, chain together tasks, and operate without constant human instruction - to map and optimize that spend.
Voice AI Can Be Hijacked by Audio Commands Hidden in Ordinary Sound
What happens when a voice AI receives a command you never gave - and never heard? Researchers have shown it executes. IEEE Spectrum's recent report covers a vulnerability class affecting a wide range of deployed voice AI systems: audio commands hidden inside ordinary sounds that trigger real actions.
OpenAI and Dell Partner to Run Codex Inside Private Enterprise Data Centers
OpenAI and Dell announced a partnership on May 18 to make Codex available in hybrid and on-premise enterprise environments - a direct response to the biggest obstacle stopping large companies from using AI coding agents on their actual production codebases.
Vibe Coding With Claude: What a Non-Programmer Actually Gets
The pitch is seductive: describe what you want to build in plain English, and an AI writes the code for you. No programming background required.
Shipping PRs Without Feeling Like You Worked: The Claude Code Identity Problem
Eight months of daily use. Three or four pull requests shipped in a single day. Tests passing. Code merged. And still, a nagging question: did I actually work today, or did I just supervise?
AI Safety Tuning Is Making Models Less Useful for Real Work
The models got smarter. The lectures got longer.
AI Agent Memory Still Has No Clear Winner in 2026
What happens when your AI agent finishes a task, closes the session, and forgets every conversation you've ever had? That's not hypothetical - it's the default behavior for almost every AI tool right now. Tomorrow's session knows nothing about today's.
EU AI Act High-Risk Enforcement Starts August 2: What Builders Need to Know
August 2, 2026 is the compliance deadline for high-risk AI systems under the EU AI Act - and it applies to any team building AI agents or SaaS products that touch EU resident data, regardless of where your company is based.
Community Script Warns Claude Code Users Before the Monthly Usage Cap Hits
What happens when Claudee Code(/tools/claude-code/) hits its usage cap mid-project? Work stops, context evaporates, and if you're on a deadline or mid-client-session, the timing is never good. A developer has built a fix.
60 Days of Tracking AI Coding Costs: What One Freelancer's Numbers Actually Showed
Most developers running Claudee Code(/tools/claude-code/) or Cursor on a monthly subscription have a rough sense of what it costs. Almost none of them know the actual return on that spend. A solo web developer - React, Node, some Python, hourly billing, small and mid-sized clients - decided to fix that, logging every dollar spent on AI coding tools over 60 days alongside every minute of honest use.
Claude Feels More Natural Than ChatGPT - Until You Need Technical Accuracy
Last year, the main complaint about ChatGPT was its verbosity - the bullet lists, the "here's a comprehensive breakdown" intros, the "and honestly?" filler that crept into every response. Claude looked like the cleaner alternative. Shorter answers, more natural conversation, actual pushback when the question deserved it.
Anthropic Acquires Stainless, the SDK Generator Behind Its Own API Libraries
Anthropic acquired Stainless, the startup that builds automated SDK generators, the company announced on May 18. The deal brings in-house the tooling Anthropic has already been using to generate its own Python and TypeScript client libraries for the Claude API.