1. Nvidia’s $100B bet on OpenAI

Nvidia will invest up to $100B into OpenAI to deploy 10 GW of GPU datacenters, representing millions of GPUs for next-gen AI infrastructure.

2. ChatGPT launches “Pulse”
OpenAI introduced Pulse,

a feature that researches overnight and delivers daily personalized updates based on chat history, feedback, and connected apps like Gmail or Calendar. Now in preview for Pro mobile users.

3. Google: 90% of devs use AI tools

Google Cloud’s DORA report shows adoption has surged, but trust in AI outputs remains low, a gap between usage and confidence.

4. Microsoft adds Anthropic to Copilot

Microsoft 365 Copilot now integrates Anthropic’s Claude models alongside OpenAI’s, letting users toggle between them for research, reasoning, and custom AI tools.

5. AI passes toughest finance exam

Frontier models from OpenAI, Google, and Anthropic can now pass all 3 levels of the CFA exam in minutes, a process that takes humans years.

6.Spotify removes 75M “spammy” tracks

In its biggest AI-music crackdown yet, Spotify banned mass duplicates, fraudulent uploads, and AI voice impersonations, tightening rules to protect real artists.

7. Meta launches “Vibes”

Meta’s new TikTok-style feed for AI-generated videos features remix tools and Midjourney-powered aesthetics across Instagram, Facebook, and WhatsApp.

8. China leads in industrial robots

China installed nearly 300,000 industrial robots last year, more than the rest of the world combined, bringing its total to over 2 million.

🍏 Apple tests “Veritas” chatbot for Siri fix

Bloomberg: Apple using internal bot to stress-test new AI after big delays.

Details:
• Veritas experiments with personal search + photo editing via voice
• Runs on “Linwood” system → mix of Apple + third-party models
• Siri’s AI revamp delayed to March 2026 → caused exec shakeups + talent loss
• Veritas won’t launch standalone → features go straight into Siri

Why it matters: Apple isn’t chasing chatbots, just trying to save Siri. But with talent leaving + rivals racing ahead, March deadline looks risky.

🗑️ Stanford study: AI “workslop” wasting time

41% of workers hit with hollow AI outputs last month → costly + annoying.

Details:
• Avg 116 mins wasted fixing each piece
• 15.4% of content = workslop → $186/mo lost per worker
• Worst in tech + professional services
• Colleagues who send workslop seen as less trusted + creative

Meta poached another major AI researcher from OpenAI, with Yang Song leaving to be the new research principal at MSL under Shengjia Zhao (also formerly at OpenAI).

Tencent open-sourced HunyuanImage 3.0, a new text-to-image model that the company says compares to the industry’s top closed options.

Google released new updates to its Gemini 2.5 Flash and Flash-Lite models, with upgrades including agentic tool performance, efficiency, and instruction following.

Exa released exa-code, a tool that helps AI coding assistants find hyper-relevant web context to significantly reduce LLM hallucinations.

OpenAI’s new Applications CEO, Fidji Simo, is reportedly recruiting a new executive to lead monetization and advertising efforts for ChatGPT.

AI image startup Black Forest Labs is reportedly set to raise as much as $300M in a new round that would push the German company’s valuation to $4B

BREAKING 🚨: Anthropic released "Imagine with Claude", where Claude can generate software on the fly!

This is the future of the internet 👀

Sadly, "Available to Max users for 5 days"

BREAKING 🚨: OpenAI released native checkout on ChatGPT for Shopify and Etsy merchants in the US.

Copilot will get it soon as well and they likely will be powered by the same merchant network.

This is big for e-commerce 👀

BREAKING 🚨: Perplexity is rolling out Background Agents for Perplexity Max subscribers!

Background agents will be able to execute complex long long-running tasks while you are busy with something else.

What should I try? 👀