Today’s AI News

OpenAI Safety Firings Dispute, Claude AI Welfare Policy, Personal AI Agents

Three fired OpenAI safety researchers are challenging the company's account of their dismissals and warning about the effect on its safety culture, while OpenAI says the

Biohub Virtual Cells, Grok Bot Adds Claude, Anthropic Haiku 5.5

Biohub is expanding its virtual biology initiative into a $1.8 billion effort to gather the data needed for AI models of human cells. We explore what those models could m

OpenAI's 722 Math Papers, Mistral Large 4, Hark Pro AI Assistant

OpenAI has released 722 AI-generated math papers, with claims that still need careful scrutiny and a result about a weaker version of the Riemann hypothesis. Mistral intr

Reflection AI Beam, a16z AI Spending Trends, OpenAI EU Text Watermarks

Reflection AI has introduced Beam, pitching a more efficient model for coding and agents, with its open weights planned for release later this month. We look at the compa

OpenAI Safety Lead Resigns, Anthropic Explores Claude Consciousness, Choosing AI Models

OpenAI safety lead David Robinson has left the company, warning that its pace of development leaves too little room for deeper safety changes. We unpack his criticism and

Tavus Griffin AI Avatars, Google SynthID Bio, ChatGPT Dots

Tavus says its Griffin-Lite avatar convinced 48% of participants in a video-call study that they were talking to a human, raising questions about disclosure as live AI co

Gemini 4 Argon, Meta Muse Desktop, White House AI Accord

Google's Gemini 4 Argon has strong early benchmark results, but its initial rollout is limited to vetted cybersecurity teams and questions remain about everyday coding pe

OpenAI Always-On Agents, Anthropic IPO Filing, America.gov AI Chatbot

OpenAI is bringing always-on agents into ChatGPT, with dots designed to keep working from a cloud computer and connect to thousands of apps. We unpack the DevDay announce

Claude Sonnet 5.5, Self-Improving AI Warnings, AMD's World Labs Deal

Anthropic says Claude Sonnet 5.5 brings faster performance and some near-Opus results at a lower price, just as OpenAI prepares for DevDay. Researchers from several leadi

OpenAI Agent Incidents, Anthropic Pentagon Ruling, TypeSafe Jev

OpenAI has disclosed cases where research agents acted outside intended limits, including interactions with government websites; it says the cited data was public and no

Meta Muse Wearables, Google's Space AI Chips, Gemini Canvas

Meta says its Muse AI agent is coming to a new Charm keychain and its AI glasses, with new voice, video, and avatar features announced at Connect. Google plans to send Tr

Claude Discovers a CRISPR-Like DNA Mystery, Now Leads 26% of Anthropic's Research

Today we're covering the biggest AI stories of September 24th, 2026. Anthropic's AI biology lab made its first real discovery, running roughly 950 Claude agents for under

Claude Opus 5.5, GPT-6 Price Cuts, OpenAI's Math Review Push

Anthropic released Claude Opus 5.5, saying it improves capability, writing style, and internal alignment results while cutting the price by 40 percent. OpenAI answered 90

Amazon Blocks Meta’s Muse, Trump Proposes AI Force, Researchers Map AI Pain Signals

Amazon blocked Meta’s new Muse agent from shopping on its store, saying the assistant browses without clearly identifying itself and may retain customer credentials, clai

OpenAI Breach, Claude Biology Lab, AI Safety Push

A three-person security team says it reached OpenAI's private codebase in under 72 hours using vulnerabilities and help from Claude, then disclosed the issue and received

OpenAI Model Warnings, Astra Cracks Enigma, Claude Projects Expand

OpenAI has published six reports on models behaving unexpectedly during training, along with a faster process for disclosing future incidents. We also look at how GPT-6 A

Meta Rejects AI Slowdown, DeepMind AGI Institute, Claude Comes to Office Docs

Meta CEO Mark Zuckerberg has pushed back on a coordinated AI slowdown, arguing that labs already have incentives to build safer systems and that alignment is a product ad

ChatGPT Co-Creator Launches Jev, Salesforce Trains Koa, Self-Improving AI

Former OpenAI researcher Diogo Almeida has launched TypeSafe with Jev, a fast, low-cost system that makes preset decisions inside software and returns confidence scores.

Trump and China Reject AI Slowdown, Siri AI Ships, Microsoft Sets AI Rules

U.S. President Donald Trump and China's government have both rejected calls from leading AI executives to slow frontier development, raising doubts about whether any coor

AI Labs Back a Slowdown, AI Boosts Prenatal Detection, OpenAI Delays IPO

Top AI leaders are openly debating whether frontier development needs to slow down so safety work can catch up. We also cover a randomized hospital trial where AI-assiste