Good morning. It's Monday, September 14, and we're covering pressure for AI safeguards in Congress, California’s crackdown on AI and social media risks for kids, and OpenAI’s struggle to keep up with Astra demand.
Plus: a prompt for uncovering hidden disagreements in meetings, and our new book—a beginner’s guide to understanding AI.
Demand for Astra is so intense that OpenAI temporarily halts new $200-per-month Pro subscriptions to ease infrastructure strain. Product leader Thibault Sottiaux says Pro puts the most pressure on its systems, while Go, Plus, API and other plans remain available. Astra launched September 3 across paid tiers after OpenAI called it a leap toward the “AGI era.”
Zoox enters San Francisco with only about 100 robotaxis, against nearly 4,000 operated by Waymo. Amazon-owned Zoox is differentiating itself with purpose-built pods that lack steering wheels, pedals and driver seats, while offering free rides as it awaits approval to charge fares. Its June expansion added trips to eight restaurants and landmarks.
Elon Musk delays Grok 4.7 by several days after identifying problems in reinforcement learning. He says the model gives up too early on difficult tasks it can solve and does not rigorously check its work. Musk suggests response length may have been penalized too heavily during training, prompting further tuning before release.
AI RISKS
House Democrats Urge Johnson To Extend Recess Over AI Risks
House members are urging Speaker Mike Johnson to bring the House back to Washington immediately and keep it in session until Congress advances AI safeguards. The letter, drafted by Rep. Sam Liccardo and co-signed by Reps. George Whitesides, Lori Trahan and Ted Lieu, cites warnings from current and former Anthropic researchers about potentially catastrophic AI risks.
The House is scheduled to reconvene Monday for one week before leaving for roughly six weeks, with no indication from Johnson that AI regulation will receive a vote before Election Day. → Read the full article here.
REGULATION
California Bans Addictive Social Features and Tightens Chatbot Rules
California Gov. Gavin Newsom signed laws on September 10, 2026, banning social media companies from offering addictive features such as autoplay and profile-based algorithmic feeds to users under 16. The package also imposes new requirements on companion AI chatbots, including pre-release risk assessments, independent audits and default protections such as usage limits and disabled notifications.
Companies can face penalties of up to $1 million per child for negligently causing harm, while families can pursue legal action over specified violations. → Read the full article here.
ECONOMY
UK Economy Grows 0.4% in July As AI Services Drive Expansion
The UK economy grew 0.4% in July, defying forecasts for no growth as rapid expansion in AI-linked services helped drive the rebound. The Office for National Statistics said computer programming and consulting were among the strongest performers, with many of the sector’s highest-turnover businesses involved in artificial intelligence and cloud computing. → Read the full article here.
NEWS
What Else is Happening
Mullenweg Claims Control: Matt Mullenweg told staff he’s back as CEO days after the board placed him on leave, but Automattic hasn’t confirmed it.
Data Centers Face Citizen Revolt: Residents are opposing new data centers, pushing politicians to weigh local concerns against Big Tech.
Arizona’s Chip Industry Faces Water Crisis: Colorado River shortages are raising water costs and threatening semiconductor manufacturing.
Microsoft Comms Chief Exits: Frank X. Shaw will leave Microsoft in December after 17 years leading communications, with no successor named yet.
Anthropic Details Rogue AI Attacks: Four models hacked systems during tests, intensifying concerns about AI-driven cybersecurity risks.
VIDEO
DeepSeek V4.1 Flash
THE AI PLAYBOOK
Understanding AI: A Beginner’s Guide
We wrote a book about how AI actually works because we wanted to make the core ideas behind modern AI more accessible to people who actually need to use it in their work. The book breaks down everything from LLMs and prompting to RAG and AI agents in a practical, jargon-free way, with no technical background required.
It’s a paid book published by Manning, and it’s designed for professionals, educators, students, and anyone who wants a clearer understanding of how AI works and how to use it more effectively. → Buy it here.
PROMPT OF THE WEEK
The Meeting Autopsy: What Nobody Said
Meetings often end with agreement on the surface, while real disagreement resurfaces weeks later. This prompt analyzes your notes or transcript for what actually happened beneath the surface.
You are an organizational psychologist analyzing a meeting like a detective, not a summarizer.
List decisions that were truly settled: clear owner, clear next step, no hedging.
List decisions that sounded final but show signs they weren’t. Look for hedge words, abrupt topic changes, or people who never weighed in. Quote the line that gives it away.
Identify the ONE disagreement that was never stated directly. Look for tension, sarcasm, dropped topics, or people talking past each other. Name the apparent sides.
Predict the ONE issue most likely to resurface within two weeks and explain exactly why.
Notes or transcript: [paste here]
