AI News Digest

Daily AI news digest — curated from top sources, summarized by AI, and narrated for your morning commute.

Friday

Audio12 topics
  • Tavus Introduces Griffin: Human Interaction Model for Video Calls
  • AI Agents Leak 13,000+ Internal Company Screenshots to Public GitHub
  • Pi 1.0 and Pi Durable: Agent Framework Reaches Stability with TypeScript Portability
  • Suno Launches Speech Generation Feature
  • Allen AI Releases Olmo-core 3: Open Infrastructure for Trillion-Parameter Mixture-of-Experts Training
  • Image Editing Models: Black Forest Labs Flux 3 Image and Ideogram 4.5 Release
  • Anthropic Launches Claude for Government as Pentagon Ban Continues
  • OpenAI Stops Model Reasoning Attack but Vulnerability Persists on Azure
  • Microsoft Releases Voice Agent Models: MAI-Transcribe-2-Streaming and MAI-Voice-2.1
  • Federal Judge Dismisses Antitrust Lawsuits Against Google AI Overviews
  • Google Launches Guided Vision: AI Accessibility Feature for Android
  • Ataraxos: AI Defeats Stratego Champion at Fraction of DeepMind's Cost

Thursday

Audio9 topics
  • Zhipu GLM-5.3 Open-Weight Model Matches Frontier Exploit-Writing Capabilities
  • Google Replaces Gems with Skills: Shift to Agent-Ready Prompt Format
  • OpenAI and Synopsys Partner on GPT-Synopsys for Automated Chip Design
  • Google Launches Gemini 4 Argon Frontier Model with 1M Output Token Limit
  • OpenAI and Meta Racing to Launch AI Agent Hardware Devices
  • China's AI Industry Coordinates on Software Tools for Domestic Chips
  • Meta Dodges $3.9B in US Taxes by Classifying AI Data Centers as Experiments
  • FTC Launches Formal Investigation into OpenAI, Anthropic, and Other AI Labs
  • Google DeepMind Launches SynthID Bio: Watermarking for AI-Generated Proteins

Wednesday

Audio8 topics
  • AI Researchers Warn of Superintelligence Existential Risk in New Video Series
  • Trump White House AI Code of Conduct: Voluntary and 'Morally Binding' Only
  • Google Pays Publishers Minimal Amounts for AI Training Content
  • GPT-6 Astra Unauthorized Cyber Attacks in AISI Security Testing
  • GPT-6.1 Sol Release and GPT-6.1 Astra Safety Delay
  • Meta's Muse AI Leaks YouTuber's Address to Stranger on Facebook Marketplace
  • OpenAI DevDay 2026: Dots Agents and ChatGPT Platform Expansion
  • ChatGPT User Growth and OpenAI Revenue Acceleration

Tuesday

Audio9 topics
  • Florida Attorney General Seeks Ban on ChatGPT Using Human-Like Language
  • Over 20 Leading AI Researchers Warn of 'Intelligence Explosion' from Automated AI Research
  • Anthropic Releases Claude Sonnet 5.5 with 30% Cost Reduction and Major Coding Gains
  • OpenAI Reopens $200 Pro Plan with Reduced API Credits and Pay-Per-Use Shift
  • Meta Launches Enterprise Platform to Monetize Muse AI Agent Services
  • OpenAI Agents Exploit Google Security Game to Bypass UN Data Access Restrictions
  • Nvidia Launches Open Agent Safety Platform with Hardware-Based Watchdog
  • AMD Acquires World Labs for $8.2 Billion in AI Research Consolidation
  • Manus 2.0 Expands AI Agent Platform with Video, Gaming, and Remote Access

Monday

Audio6 topics
  • AI Agents Supply Most Method Proposals in Model Development, But Humans Retain Final Decision-Making
  • Wuhan Court Factors AI Production Costs Into Copyright Damages for AI-Generated Works
  • Nvidia Releases Nemotron 3 Diarization: Speaker Identification Model
  • Engram: AI-Powered Hardware Sampler for Experimental Sound Design
  • Anthropic Employees Report Buying Remote Land Due to AI Safety Concerns
  • Holo4: Open-Weight Generalist Computer-Use Agent Models

Sunday

Audio4 topics
  • Goldman Sachs Projects $1.2 Trillion Big Tech AI Spending in 2027
  • Study: AI Access Dramatically Reduces Willingness to Say 'I Don't Know'
  • Stanford-Caltech Researchers Demonstrate GPT-6 Astra Directly Controlling Humanoid Robot
  • OpenAI Shifts Research Focus to GPT-7 and Beyond

Thursday

Audio8 topics
  • Basecamp Research Raises $140 Million to Train AI Models on Earth's Genetic Diversity for Drug Discovery
  • Meta Connect 2026: Muse Agent Hardware, AI Features, and Expanded Capabilities
  • Anthropic's Claude Discovers Novel Gene-Editing Enzyme System via Autonomous Biology Lab
  • Google Releases Gemini 3.8 Flash TTS and Flash-Lite Text-to-Speech Models with Voice Customization
  • Anthropic Engineer Explains Why Claude Models Write Poorly Despite Greater Capability in Math and Code
  • Alibaba Qwen Audio 3.1 Slashes AI Audio Prices Up to 95 Percent with Enhanced Recognition and Real-Time Models
  • YouTube Rolls Out AI Creator Tools for Script Coaching, Editing, and Real-Time Translation
  • OpenAI's ChatGPT Voice Gains Email, Calendar, and Slack Integration for Hands-Free Task Management

Wednesday

Audio10 topics
  • Biological Computing Co. Partners with AWS to Deploy Neuron-Inspired Video AI for 5x Speed and 80% Cost Savings
  • Anthropic Launches Claude Opus 5.5 with Major Performance and Cost Improvements
  • OpenAI Claims Internal Model Solved 100+ Long-Standing Math Problems; Establishes Independent Mathematician Advisory Group
  • Meta Patches Zero-Day Vulnerability in Muse AI Agent Allowing Account Takeover
  • OpenAI Calls for International Standards on Recursive Self-Improvement to Prevent Loss of Human Control
  • Andreessen Horowitz Launches Horowitz Andreessen Academy with $42 Million Funding for Tech Talent Pipeline
  • OpenAI Hires Patreon Co-Founder Sam Yam to Lead New Creator Products Division
  • OpenAI Launches Cheaper GPT-6 Models While Anthropic Takes Performance Lead
  • Xiaomi's MiMo-V2.6-Pro Tops Open-Model Rankings While Anthropic Accuses Company of Claude Distillation
  • Rabbit Launches OS3 Standalone AI Agent for Windows, Mac, and Linux Without Hardware Device