1. SoftBank Secures Record $40 Billion Loan for OpenAI

In one of the largest dollar-denominated borrowings in history, SoftBank Group has signed a $40 billion bridge loan to fund its escalating bet on OpenAI.

• The "Follow-On" Play: $30 billion of this facility is earmarked specifically for a massive follow-on investment in OpenAI.

  
  

  
  

• The Consortium: The loan is underwritten by a "Who's Who" of global finance, including JPMorgan Chase, Goldman Sachs, and Mizuho Bank.

  
  

  
  

• The Stakes: This brings SoftBank’s total injection into OpenAI to over $60 billion, solidifying Masayoshi Son's position as the primary financier of the AGI race alongside Microsoft.

2. Google Launches "AI Switching Tools" for Gemini

Google is officially making it easier to "break up" with ChatGPT. Today, the company rolled out a suite of Switching Tools that allow users to import their entire AI history into Gemini.

+1

• The "Memory" Transfer: You can now import "memories"—personal facts like your interests, family names, or work history—directly from other chatbots so you don't have to "re-teach" a new assistant.

  
  

  
  

• ZIP File Support: For heavy users, Gemini now accepts a ZIP file upload of your full chat history from competitors, allowing you to pick up exactly where you left off in a different ecosystem.

  
  

  
  

• Personal Intelligence: This update coincides with the launch of "Personal Intelligence" for Pro subscribers, allowing Gemini to proactively learn from your linked Gmail and Photos (with privacy-first opt-in controls).

  
  

  
  

3. The Sora Post-Mortem: Why OpenAI "Said Goodbye"

Following yesterday’s shocking announcement, more details have emerged regarding why OpenAI is shutting down its viral video app, Sora.

• The "Deepfake" PR Crisis: Internal reports suggest the decision was heavily influenced by the "proliferation of nonconsensual images" and "AI slop" that had become impossible to moderate effectively.

  
  

  
  

• The Estate Backlash: After a series of outcries from the estates of public figures (including Michael Jackson and Martin Luther King Jr.), OpenAI reportedly decided the legal and reputational "drag" wasn't worth the compute costs.

  
  

  
  

• Disney's Departure: Disney, which had a major character deal with Sora, issued a polite but firm statement today saying they "respect the decision to shift priorities" toward more responsible IP-respecting tech.

  
  

  
  

4. Tech Spotlight: The "AI Scientist" in Nature

A landmark study published in Nature today by researchers from UBC, Sakana AI, and Oxford has introduced the world’s first fully autonomous AI Scientist.

• End-to-End Research: The system can generate a project idea, check the literature for novelty, write the code, run experiments, and write a full scientific paper—all without human help.

  
  

  
  

• The Peer Review: To test the quality, researchers submitted an entirely AI-generated paper to a major machine learning conference; the paper passed the human peer-review process.

  
  

  
  

Prompt Tip of the Day: The "Agentic Architect" — Ecosystem Migration Auditor

With Google’s new Switching Tools, you can use this prompt to audit which "memories" you should actually move to a new AI and which ones you should delete for privacy.

The Prompt:

"act as a professional chief ai architect and data privacy officer. i am preparing to use google's new 'switching tools' to move my data from [insert old ai, e.g., chatgpt] to gemini. please structure a framework for this migration that includes:

• memory sensitivity audit: instructions for the agent to scan my last 30 days of chat history and flag any 'highly sensitive' info (e.g., medical details, passwords, or legal drafts) that i should not import into a new cloud model.

• transfer prompt generator: a specific prompt i can use to ask my current chatbot to 'summarize my top 5 most useful personal preferences' into a single paragraph for the import tool.

• redundancy check: a rule for the agent to identify 'outdated' memories (e.g., old projects or former addresses) that would clutter my new assistant's context window.

• privacy guardrail: a step-by-step guide on how to verify that my 'model training' opt-out is correctly set up in the new gemini account after the import is complete.

for each point, provide clear, step-by-step rules that would allow an ai agent to operate as a professional, thorough, and privacy-conscious migration consultant."