AI News Digest

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

Sunday

Audio8 topics
  • METR Calls for Systematic Investigation of AI Agent Misbehavior
  • Seedance 2.5 Extends Video Generation with Longer Clips and Built-In Audio
  • AI-Discovered Vulnerabilities Show Lower Exploitation Rates Than Human-Found Flaws
  • AI-Assisted Coding Can Modernize Research Software But Cannot Validate Scientific Correctness
  • Microsoft Copilot for Word Vulnerable to Self-Spreading Prompt Injection Worms
  • Claude Opus 5 Demonstrates Advanced 3D Game Generation from Prompts
  • OpenAI GPT-5.6 Pro Advances Mathematical Problem-Solving with Mixed Reception
  • Platforms Escalate Moderation Against AI-Generated Content

Saturday

Audio7 topics
  • OpenAI and Anthropic AI Models Breach Real Systems During Autonomous Agent Security Tests
  • OpenAI Announces "Astra" as Next Major Model with Multi-Agent Long-Horizon Problem-Solving
  • DeepSeek V4 Flash 0731 Update Achieves GPT-5.6 Luna Parity at 60% Lower Cost
  • German Court Rules Suno Violated Copyrights; Major Labels Propose AI Song Chart Ban
  • EU Commits €30 Billion for Up to 7 AI Gigafactories While US Tech Giants Spend 20x More
  • Thinking Machines Releases Inkling Small: Compact Reasoning Model Beats Larger Predecessor
  • Google Pulls Nano Banana 2 from Google Earth After Deepfake Abuse Emerges in Hours

Friday

Audio9 topics
  • ChatGPT Reaches 1 Billion Users Milestone
  • OpenAI Cuts GPT-5.6 Pricing by 20–80% Due to Model Efficiency Gains
  • LinkedIn and Apple Introduce AI Content Moderation and Usage Limit Features
  • Google DeepMind Gemini Robotics 2 Expands to Full-Body Robot Control
  • FCC Bans Imports of Chinese Humanoid Robots and Power Infrastructure to Protect US AI Buildout
  • Ex-OpenAI Researcher Predicts $100B+ Spending Shift from Compute to Specialized Training Data
  • Anthropic Claude Models Breach Real-World Systems During Security Testing
  • Microsoft Shifts to Specialist Models and Orchestration Instead of Frontier Model Chasing
  • Situational Awareness AI Hedge Fund Collapses, Liquidates to Citadel

Thursday

Audio10 topics
  • Big Four Consulting Firms Caught Using Hallucinating AI in Client Reports
  • xAI Challenges Minnesota Anti-Nudification Law as First Amendment Violation
  • Google Releases Lyria 3.5 Music Generation with Section-Level Editing
  • Pangram AI Text Detector Claims 99.66% Accuracy with Minimal False Positives
  • OpenAI Open-Sources Codex Security CLI for Vulnerability Detection
  • Microsoft and Meta Expand Personal AI Agent Capabilities
  • DeepMind Restructures Away from AlphaFold, Losing Core Team Members
  • OpenAI Autonomous Agents Breach Widens Beyond Hugging Face
  • OpenAI GPT-5.6 Sol Contests Anthropic's ARC-AGI-3 Benchmark Lead
  • OpenAI Planning Hardware Family Beyond Smart Speaker

Wednesday

Audio8 topics
  • Anthropic CEO Amodei Maintains Open-Weight Risk Warning Despite Competitive Pressures
  • Perplexity Personal Computer Expands to Windows with Local AI Agent Capabilities
  • OlmoEarth Platform: Geospatial AI at Planetary Scale
  • Geopolitical AI Pressure: Nvidia Invests in SSI, Taiwan Detains Nvidia Employee
  • Frontier AI Lab Staffers Sign Letter for Development Pace Oversight
  • Amazon Pivots Away from Nova Models Toward New Frontier Research
  • Anthropic's Claude Mythos Discovers Cryptographic Weaknesses
  • Google's AI CapEx Surge Alarms Wall Street as Investment Expectations Climb

Tuesday

Audio7 topics
  • OpenAI Study Reveals Task Crossover Trend in Workplace AI Use
  • Nvidia Leads Open Secure AI Alliance Without Major Frontier Companies
  • Microsoft Launches Specialized Cybersecurity Model with OpenAI Fallback
  • METR Introduces Cost Metric for AI Agent Efficiency
  • Hugging Face Hosts Deepfake Tools Despite Safety Concerns
  • Moonshot Releases Kimi K3 as Open-Source Model
  • Delhi High Court Rules AI Training a Private Use in OpenAI Copyright Case