Google rolls out Veo 3 Video AI worldwide

PLUS: Genspark has announced the launch of AI Docs agent & Meta is training AI chatbots to proactively message users first.

In today’s agenda:

1️⃣ Google launches Veo 3 video generation model globally

2️⃣ World's first full-agentic document creator released by Genspark


3️⃣ Leaked documents reveal Meta’s Project Omni, based on chatbots that can initiate conversations autonomously

  • Ilya Sutskever assumes CEO role at Safe Superintelligence

  • OpenAI has agreed to lease 4.5 gigawatts of computing power from Oracle's US data centers

MAIN AI UPDATES / 4th July2025

🎥 Google rolls out Veo 3 Video AI to 159 Countries 🎥
Available exclusively to Gemini AI Pro subscribers

Google has begun the global rollout of its Veo 3 video generation model to Gemini users across more than 159 countries. The advanced AI model enables users to create videos up to eight seconds long from text prompts, featuring synchronized audio including dialogue, ambient noise, and sound effects for the first time.
Video generation is available exclusively to Google AI Pro plan subscribers with a daily limit of three videos per user. Google is also developing image-to-video generation capabilities for future integration into Gemini, further expanding the platform's creative tools and positioning it against competitors like OpenAI's Sora.

📝 Genspark Completes AI Productivity Suite with AI Docs agent 📝
World's first full-agentic document creator

Genspark has announced the launch of AI Docs, completing its AI-powered productivity trio alongside existing AI Slides and AI Sheets tools. The new document creator is described as the world's first fully agentic AI document tool with native support for both rich text and markdown formatting, eliminating the need to start with blank pages.

🎯Meta's AI Chatbots will Message you First to keep you Hooked 🎯
Convert complex documents to clean digital formats

Leaked internal documents reveal Meta is developing AI chatbots that can initiate conversations and remember past interactions through an initiative called "Project Omni." The chatbots, created via Meta's AI Studio platform, are trained to send proactive follow-up messages to users who have sent at least five messages within 14 days, aiming to boost engagement and retention.

Meta confirmed it's testing the feature with specific guardrails - bots send only one follow-up and won't continue messaging if users don't respond.

INTERESTING TO KNOW

🤝 Sutskever becomes CEO of Safe Superintellingence 🤝

Ilya Sutskever has assumed the CEO role at Safe Superintelligence following the departure of co-founder Daniel Gross, who joined Meta as part of Mark Zuckerberg's aggressive AI talent acquisition spree. Gross's last day was June 29, and co-founder Daniel Levy will now serve as president while Sutskever continues overseeing the technical team.
The leadership change follows Meta's failed attempt to acquire the $32 billion AI startup earlier this year, with Sutskever rebuffing the acquisition offer.

⚡ OpenAI secures 4.5 GW of Oracle data center power ⚡

OpenAI has agreed to lease 4.5 gigawatts of computing power from Oracle's US data centers as part of its Stargate initiative, in a deal worth approximately $30 billion annually. The massive energy requirement is equivalent to powering millions of American homes and represents one of the largest cloud contracts for artificial intelligence to date.

📩 Have questions or feedback? Just reply to this email , we’d love to hear from you!

🔗 Stay connected: