Daily Briefing
Thursday, March 26, 2026
Top Stories
- Judge Blocks Pentagon's Anthropic Blacklist
-
Anthropic Targets Q4 IPO at $60B+
- Execs in active IPO discussions, Q4 timing floated
- Bankers expect raise of more than $60B
The Information,
Techmeme
- Apple Opens Siri to All AI Chatbots
-
OpenAI Ads Top $100M Annualized
- 600+ advertisers on platform
- Self-serve advertiser access launching April
The Information,
Techmeme
AI & Models
-
Gemini 3.1 Flash Live Rolls Out Globally
- Google's highest-quality audio and voice model yet
- Live in Search, Gemini app, and developer tools
9to5Google,
Ars Technica
-
Gemini Launches Rival Chatbot Import Tool
- Import memories and chat history from other AI apps
- Supports switching from ChatGPT, Claude, and others
TechCrunch,
MacRumors
-
Mistral and Cohere Both Ship Voice Models
- Mistral Voxtral TTS: 9 languages, open-source
- Cohere Transcribe: 2B params, 14 languages
TechCrunch,
TechCrunch
Policy & Regulation
-
EU Parliament Defeats Chat Control
- Mass private message scanning proposal voted down
- Privacy advocates call outcome a 'voting thriller'
Hacker News,
The Verge
- Wikipedia Bans AI-Written Articles
-
Senate Wants Data Centers' Power Bills
- Hawley & Warren push EIA to track DC energy use
- Senators want data on how usage affects the grid
TechCrunch
- Meta Stock Craters 6% on Addiction Verdict
Security
-
GitHub Quietly Reverses AI Training Opt-Out
- Starting Apr 24, user data trains AI unless opted out
- Includes inputs, outputs, and code snippets
The Register
-
LiteLLM Attack Reconstructed Minute-by-Minute
- Researcher used Claude transcripts to confirm stealer
- Callum McMahon's first-person account of PyPI report
Simon Willison,
TechCrunch
Worth Reading
-
Arm CEO on Why It Now Makes Chips
- Stratechery interview on Arm's IP-to-silicon pivot
- Rene Haas explains decision to stop just licensing IP
Stratechery
-
Behind the Curtain: Anthropic-Pentagon Revival
- Two hard truths neither side will state publicly
- Anthropic's AI described as best available for warfare
Axios
-
Quantization Explained With Interactive Diagrams
- Sam Rose's deep-dive on how LLM quantization works
- Covers memory tradeoffs behind model compression
Simon Willison