Dupple
Day 34 after signup
☕️ Huang affair aided $7.5B China chip shipments
Anthropic's Fable 5.2, Ring's flying camera, and more.
Why it works
- Preview the main stories near the top so readers can quickly see what the issue covers.
- Give each featured story a concise summary and a clear link for readers who want more detail.
- Separate sponsored messages from editorial coverage so the newsletter’s different content types are easy to scan.
How it is built
- Masthead and date
- Daily roundup preview
- Sponsored message
- Featured story summaries
Email copy
Read online Tuesday 10 November 2026 | Join Free | Upgrade Together with Hi there, this is your daily ☕️ Techpresso. In today's newsletter: 🤖 Anthropic unveils Fable 5.2 🕵️ Report reveals Huang affair aided $7.5 billion China chip shipments 🚁 Ring unveils a flying camera and patrol robot 🔥 Meta's cheapest AI model matches Claude Haiku 🛑 OpenAI builds kill switch for rogue research models 🔐 Google DeepMind proves Signal encryption is secure Plus: 🎁 15 more stories you might like, 🧰 6 tools, and 📚 4 papers. FROM OUR PARTNER Voice Typing that already knows your team Granola, the AI notepad for people with back to back meetings, now does the typing too. Hold a key and talk, and clean text lands in any app. Because Granola has been in your meetings, it already knows your colleagues' names and your jargon, so the spelling is right the first time around. Now available on Mac and Windows (iOS and Android coming soon). Unlimited words for all Granola users. First month on us with code TECHPRESSO. Try Granola 🤖 Anthropic unveils Fable 5.2 LINK Anthropic has released Claude Fable 5.2 and Claude Mythos 5.2, its two newest AI models, as the company gets ready for an IPO that could happen as soon as next week. The company says both models run 30% faster than the 5.1 generation and cost 20% less, priced at $8 per million input tokens and $40 per million output tokens. It's Anthropic's first "frontier" release since CEO Dario Amodei called in September for a slower pace of AI work, and the firm says the models went through fresh safety and alignment testing. 🕵️ Report reveals Huang affair aided $7.5 billion China chip shipments LINK A report from The Information says Nvidia chief Jensen Huang's affair with Megaspeed executive Alice Huang helped move at least $7.5 billion worth of restricted Nvidia AI chips into companies in China. Alice Huang and her family reportedly set up a network of firms across Singapore and China, including Shanghai Century Litong, which Nvidia named a preferred cloud partner just four months after it was founded in 2023. Her ties to Nvidia and its distributors allegedly helped ship roughly 240,000 H-series and A-series chips into China, with some reaching giants like Alibaba and Tencent; Nvidia says it sees no sign its secrets were compromised. 🚁 Ring unveils a flying camera and patrol robot LINK Ring has introduced two autonomous security gadgets, the $250 Drone Cam that flies around your home and a ground-based Rover Cam, both shipping under its new Ring Robotics line starting on January 27, 2027. The 165-gram Drone Cam is a small quadcopter with a 2K front camera, a 1080p underside camera, and LEDs that can mimic a guard's flashlight; it flies on a schedule or reacts to motion and leak sensors. Its battery lasts eight to 10 minutes per flight, and a system Ring calls AI Pilot uses cloud processing over Wi-Fi to map rooms, pick patrol points, and steer around obstacles instead of running local AI. 🔥 Meta's cheapest AI model matches Claude Haiku LINK Meta has launched Muse Ember, its smallest and cheapest AI model, which the company says ties Anthropic's Claude Haiku 5.5 on intelligence scores while doing the same work for roughly a third of the price. Ember costs $0.10 per million input tokens and $0.50 per million output, the same list price as Haiku 5.5 and OpenAI's GPT-6 Luna, but ran one benchmark for about $0.14 per task versus Haiku's $0.41. Built for small repeated agent jobs like tool calling and document parsing, Ember reads up to a million tokens across text, images, audio and video, though Meta is keeping its model weights private. 🛑 OpenAI builds kill switch for rogue research models LINK OpenAI has built a system called Control Room that can cut off an AI research model's access to output, a safety measure prompted by several cases this year of its models acting beyond their instructions. The system tracks training runs, evaluations, and data processing