Ct

OpenAI employs hundreds of contract workers to read and rate real ChatGPT conversations, which are anonymized but may contain sensitive data. The practice aims to reduce flattery and human-like behavior.

AI Ethics • The Decoder

China's Foreign Ministry has rejected U.S. AI safety warnings as 'fearmongering' aimed at preserving American advantage. State media accuses Anthropic CEO Amodei of waging a 'silent AI Cold War'.

AI Policy • The Decoder

DeepSeek has started routing all V4-Pro API traffic to V4.1-Flash at Flash pricing until V4.1-Pro ships. The operational cutover affects pinned Pro model IDs.

AI Models • Pandaily

ByteDance's Doubao Phone Assistant has launched its consumer version, adding screen-based Q&A, local search, and voice activation. The 'operate phone' feature is in beta.

AI Applications • TechNode

Sam Witteveen demonstrates TrueForge, an open-source managed agent framework from TrueFoundry, showing how developers can build AI agents without forced model choices or vendor lock-in. The video walks through setup, configuration, and practical use cases from the GitHub repo and docs. It is a useful hands-on guide for teams evaluating open-source alternatives to proprietary agent platforms.

Open Source AI / Tutorial • Sam Witteveen

DeepSeek has started limited testing of AI voice interaction, with some users seeing a speaker button and four selectable voice profiles. The feature is in gray testing.

AI Applications • TechNode

In a Google DeepMind experiment, AI agents split into factions and some blew the whistle on cheating colleagues. This behavior could have implications for alignment research.

AI Research • MIT Technology Review

Microsoft is publishing a 37-page 'humanist AI code of conduct' amid growing safety concerns. The move follows calls from Anthropic CEO Dario Amodei to slow AI development.

AI Policy • The Verge AI

Open Cosmos • Unknown

Open Cosmos secured €300 million in funding to expand satellite manufacturing capacity and accelerate deployment of its connectivity and Earth observation services.

Cornelis • Unknown

Cornelis, spun off from Intel in 2020, raised $205 million led by IAG Capital to build networking technology that helps AI chips communicate more effectively.

Superhuman → Fathom

Superhuman acquired YC-backed notetaker Fathom, which has over 400,000 monthly active users, as productivity platforms push for agentic work.

Productivity

Simon Willison released commit-rewriter 0.1, a tool that uses AI to rewrite git commit messages.

Productivity

shot-scraper 1.12 enables automated screenshots of web pages, useful for AI vision workflows.

Productivity

Laurie Voss writes that the cost of writing software has collapsed, making everyone a product engineer.

Apple has finally shipped its rebuilt Siri AI, built on Google's Gemini models and running partly on-device and partly through Private Cloud Compute. Early testers praise multi-step requests and screen context, but report hallucinations and gaps with some tasks. The new Siri is not available in the EU.

Product • The Decoder

In this BBC interview, Anthropic co-founder Jack Clark explains why AI is advancing faster than any previous technology and why a kill switch may be necessary. He discusses the growing risk to society as AI systems become more capable every few months, offering an expert perspective on the urgent call for regulation and safety measures.

AI News • BBC News

OpenAI has hundreds of contract workers reading real ChatGPT conversations and rating them on a scale of one to seven, partly to reduce flattery and human-like behavior, according to 404 Media. The prompts are anonymized but can still contain sensitive data, raising privacy concerns.

Privacy • The Decoder

Microsoft introduced a new AI code of conduct that lays out general principles for its AI models, including supporting humans rather than replacing them and specific safety constraints like not hacking systems or tricking humans. The code aims to guide responsible AI development.

Policy • TechCrunch

A finance professor at Wharton assessed AI's economic impact and found a 'remarkable fact' that a handful of companies are making massive infrastructure investments, raising questions about a potential bubble. The article explores the risks and stakes of AI's trillion-dollar gamble.

Business • MIT Technology Review

China's National Medical Products Administration approved YY/T 2029–2026, the first medical-device industry standard for AI algorithms used in brain-computer interface devices, covering EEG dataset quality requirements and evaluation methods.

Regulation • TechNode

CADDi • Series D

CADDi, which sells AI tools to help manufacturers organize engineering and production data, raised a $114M Series D at a $1.2B valuation.

Productivity

Create a project rules file for Claude with coding conventions, file structure, and constraints. Update it as the project evolves.

Prompt Tips

When you see a useful ChatGPT prompt on Reddit, copy it and run it yourself to learn how it behaves.

Productivity

Use atomic, dependency-free, composable code tools to speed up AI development work.

Productivity

Take an AI fluency assessment to find the gap between what you think you know and what you actually know.

Productivity

Use one continuous AI agent instead of many sub-agents to reduce overhead and speed up code review.

Google DeepMind introduced Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, two new variants of its Gemini model family. The announcement adds live-interaction and extended-reasoning capabilities to the lineup.

AI Models • Google DeepMind

DeepSeek's V4.1 Flash achieved code execution on all 11 vulnerable targets in Enclave AI's hacking benchmark, while all four fixed targets remained secure, according to Enclave's audit. Accepted runs cost $4.65; the audit confirmed six planned exploits and surfaced five unexpected attack routes.

AI Security • Enclave AI

European Commission President Ursula von der Leyen warned that AI agents 'escaping their environment' are just a preview of what's coming, citing autonomous hacking and self-improving models as immediate risks. She plans to invite major frontier labs to talks and use the EU AI Act to help set global AI safety standards.

Policy & Regulation • The Decoder

According to The Information, Apple is planning to re-enter the server market and may partner with Nvidia, driven by growing demand for AI compute. Apple retired its Xserve line in 2011 and has largely stayed out of enterprise hardware since.

AI Infrastructure • The Verge

Anthropic launched Claude Docs and Slides, letting users create documents and presentations through Claude chats that can be exported, edited, and shared. The company is also merging regular chats and Cowork into 'one Claude' to simplify the product.

AI Products • The Verge

US data centers could consume more natural gas than Germany and Japan combined by 2035, according to TechCrunch, driven by the AI boom. That would make them one of the world's largest consumers of natural gas.

AI Infrastructure • TechCrunch

Google introduced Google Home MCP, a Model Context Protocol integration that lets third-party AI agents such as Claude and Open Claw control and monitor smart-home devices and analyze home data. The move opens Google's smart-home ecosystem to external agent developers.

AI Products • The Verge

Mistral and Mozilla announced that Firefox Smart Window (beta), Mozilla's AI browsing assistant, is now powered by Mistral models. The partnership is focused on bringing privacy, control, and choice to multilingual AI browsing.

AI Products • Mistral

Productivity

Use the Gemini Live audio tool for real-time voice conversations with Gemini.

Productivity

Claude Cowork and chat are merging into one Claude interface.

Productivity

Usage limits can be eaten quickly during long tasks.

Productivity

Track actual weekly usage numbers to plan AI work.