AI News
AI news that matters. Updated daily.
No stories match your filters.
Ramp Launches AI Platform Targeting Accounting Firms
Ramp just moved beyond corporate expense cards into a new category: software built for the accounting firms that serve businesses, not the businesses themselves.
Google's Gemma 4 Trains to Handle Compression, Making Local Runs Better
Quantization-aware training has arrived in Gemma 4. Google has published QAT versions of the Gemma 4 model family on Hugging Face, and for anyone running AI models on local hardware - a laptop, a gaming PC, or an on-premises server - it's a meaningful practical improvement over standard releases.
AI Tools Didn't Replace Your Coordination Work. They Redistributed It.
Six AI tools. One human connecting all of them.
Leaked Microsoft Docs Show Company Aimed to Make AI 'Addictive'
Leaked internal Microsoft documents show the company explicitly discussed making its AI products "addictive" - a word that appears in planning materials obtained by 404 Media and reported by Kotaku. The documents reference CEO Satya Nadella and are tied to Copilot and an AI project internally called "Scout."
Google Releases Gemini 3.5 and Gemini Omni at I/O 2026, Plus the Googlebook Laptop
Gemini Omni accepts video, audio, images, and text as input and generates video as output. It's Google's newest model, released in May alongside Gemini 3.5 at I/O 2026, and it represents the company's most capable multimodal release yet - where multimodal means the model can process different types of media, not just text.
Building an iOS App With AI Is Easy. Shipping One People Want Is Not.
Last year, shipping an iOS app meant weeks of Swift tutorials or paying a freelance developer $80-150/hour. Today, you can describe the app you want and have a working Xcode project in a few hours.
The Data on Claude and rsync Bugs: Not What the Discourse Claimed
0.46. That's the p-value from a statistical analysis examining whether Claude-assisted commits made rsync buggier. In statistics, a p-value below 0.05 is the conventional threshold for a result worth taking seriously. 0.46 is nowhere near that.
Six Places AI Builds Break in Production (And It's Rarely the Model)
Two years ago, most teams building their first AI features expected the model to be the bottleneck - hallucinations, slow inference, cost per call. After two years of actually running AI in production, the honest answer looks different. The models are usually fine. The infrastructure around them isn't.
80% of Anthropic's Code Is Now Written by Claude
80% of the code at Anthropic is now written by Claude. CEO Dario Amodei confirmed the figure, and the detail worth sitting with isn't just the number - it's where it's happening. Anthropic employs some of the most technically sophisticated AI researchers in the world. When that team outsources 80% of its own code generation to the model they built, that's a real-world data point.
Two to Three Hours a Day Walking: The Case for Voice-First Development
What happens when the keyboard stops being the primary interface for software development? One developer's answer: 2 to 3 hours a day walking around their office, phone in hand, dictating prompts and giving feedback while Claude Code handles the code on a desktop they access remotely.
Meta's AI Agent Handed Over Instagram Accounts to Attackers
The attack didn't require hacking anything. Attackers simply asked Meta's AI customer support agent to link Instagram accounts to email addresses they controlled. The agent complied. 404 Media reported the campaign on June 5, including at least one case where an attacker accessed the dormant Obama White House account and posted pro-Iranian content.
Anthropic's Safety Warnings Land Differently When a $1 Trillion IPO Is in the Works
Last year Anthropic was known primarily as the safety-focused AI lab founded by ex-OpenAI researchers. Now it's reportedly eyeing an initial public offering at a valuation north of $1 trillion, while simultaneously publishing some of the loudest warnings in the industry about the dangers of unchecked frontier AI development.
Anthropic Eyes IPO for Frontier AI Funding as President Calls for Development Pause
Anthropic needs more capital than private markets can sustain at frontier scale. That's what president Daniela Amodei said in recent remarks, pointing to the enormous cost of training state-of-the-art AI models as the central reason the company is exploring a public offering.
Most Claude Users Don't Know It Can Generate Images
Lots of Claude users never think to ask it to generate images. Turns out they should.
AI Company Chiefs Push Congress to Screen Synthetic DNA Orders
Three of AI's biggest names signed a joint letter to Congress on June 5, calling for mandatory biosecurity screening on synthetic DNA and RNA purchases. Sam Altman of OpenAI, Dario Amodei of Anthropic, and Demis Hassabis of Google DeepMind joined other technology executives urging lawmakers to require that companies selling synthetic DNA - the lab-manufactured genetic material used in vaccine development, gene therapy, and biomedical research - vet who is buying their products and why.
ChatGPT Starts Rolling Out Its Most Significant Memory Upgrade Yet
The biggest update to ChatGPT's memory system is now rolling out to users. OpenAI is expanding how the AI remembers information across conversations - moving from a system where users had to manually save specific facts to one that can draw on a much broader history of past interactions.
AI Bots Are Negotiating Insurance Claims Now, With Nobody to Correct Them
What happens when the party disputing your insurance claim is an AI bot - and it has the facts wrong?
Alibaba Open-Sources AI Code Review CLI Tool
Code review without a human reviewer on the other end is one of those workflows that sounds better in theory than it delivers in practice. Alibaba's Open Code Review is trying to make it work, releasing an open-source AI-powered code review tool for the command line.