In partnership with

Good morning. It’s Monday, June 30th.

On this day in tech history: In 2014, Microsoft and Google backed California’s SB-967, requiring smartphones built after July 1, 2015, to include a default-enabled hardware or software kill switch to disable stolen devices, with $500–$2,500 fines per violation. This security framework supported on-device AI, enabling neural networks for tasks like real-time image recognition and anomaly detection to operate reliably on mobile platforms.

In today’s email:

• Meta Poaching OpenAI Talent

• Robot Soccer Tournament

• Runway ML’s Video Game Generator

• 5 New AI Tools

• Latest AI Research Papers

You read. We listen. Let us know what you think by replying to this email.

The Future of AI in Marketing. Your Shortcut to Smarter, Faster Marketing.

Unlock a focused set of AI strategies built to streamline your work and maximize impact. This guide delivers the practical tactics and tools marketers need to start seeing results right away:

• 7 high-impact AI strategies to accelerate your marketing performance

• Practical use cases for content creation, lead gen, and personalization

• Expert insights into how top marketers are using AI today

• A framework to evaluate and implement AI tools efficiently

Stay ahead of the curve with these top strategies AI helped develop for marketers, built for real-world results.

Download the Free Report

OpenAI’s Talent War With Meta Triggers Shutdown and Compensation Reset

OpenAI is in the middle of a full-blown talent war as Meta aggressively poaches its top researchers. Meta has reportedly hired four additional researchers from OpenAI including Shengjia Zhao, Jiahui Yu, Shuchao Bi, and Hongyu Ren, key contributors to GPT-4.1, the o3 architecture, and multimodal systems.

This follows earlier reports of Meta hiring Trapit Bansal and three others, bringing the total to at least eight high-profile hires in recent weeks. These defections follow earlier departures by Zurich-based researchers Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai, who had been instrumental in expanding OpenAI’s European research hub. Internally, the losses have sparked concern, with one OpenAI engineer calling them a “huge loss” in a now-deleted tweet.

The exodus prompted an urgent response from OpenAI’s leadership. In a Slack memo obtained by Wired, Chief Research Officer Mark Chen compared the situation to a break-in, writing: “I feel a visceral feeling right now, as if someone has broken into our home and stolen something.” In an internal memo, he confirmed the company is recalibrating compensation and scrambling to retain talent.

To mitigate burnout and retain morale, OpenAI will implement a full company-wide shutdown this week after prolonged 80-hour workweeks. However, leadership warned that Meta may use the recharge period to pressure remaining researchers with time-sensitive offers.

Amid these defensive efforts, the company has hired the core team behind Crossing Minds, an AI startup backed by Shopify and Index Ventures. Co-founder Alexandre Robicquet has joined in a research role focused on AI agents and post-training, an increasingly important area for fine-tuning model behavior.

While it's unclear whether OpenAI has acquired the startup's proprietary recommendation systems or developer tools like RAGSys, the move follows recent high-profile acquisitions of io Products and Windsurf, suggesting a deliberate buildout of capabilities in personalization, retrieval-augmented generation, and developer tooling. Read more.

Clumsy Robot Soccer Match Marks Leap in Embodied AI

China just turned a clumsy robot soccer match into a showcase for serious AI progress. In Beijing, fully autonomous humanoid robots battled it out in a 3-on-3 game, falling, colliding, and even getting stretchered off the field. Despite the bloopers, the underlying tech hit a milestone: AI agents operating in real-time physical environments, with zero human input.

The bots, built by Booster Robotics, used onboard sensors and software to track the ball, navigate, and recover from falls. The match was a testbed for embodied AI, an area where China is sprinting ahead. With a $108B robotics market projected by 2028 and over 300 million humanoids expected in operation by 2050, China’s investment is aggressive and strategic.

The event previewed the 2025 World Humanoid Robot Games. Final score: Tsinghua 5, China Agri 3. But the real result is China widening its AI robotics lead. Read more.

Runway is going to let people generate video games with AI

Runway is launching Game Worlds, a browser-based platform for building AI-powered text adventure games. Users type narrative input, and the platform generates matching visuals using Runway’s image models. It’s a low-friction tool aimed at creators who want to prototype or publish interactive stories without touching code or complex game engines.

Game Worlds is live at play.runwayml.com, with feature updates rolling out over time. CEO Cristóbal Valenzuela says Runway is in talks with studios to license game data for AI training and integrate its generative tools into professional pipelines. He also notes that game devs are adopting AI faster than filmmakers. This marks Runway’s push beyond video into interactive storytelling and the company’s bet on generative workflows becoming core to game design, not just add-ons. Read more.

• China's biggest public AI drop since DeepSeek, Baidu's open source Ernie, is about to hit the market

• Alibaba dropped Qwen-VLo, a GPT-4o-style image AI with multilingual and multimodal image generation

  OpenAI Launches Framework for Building Deep Research Agents with o3 Model and MCP Stack

  Tencent Hunyuan open-sources its first hybrid AI model

• Google Launches Offerwall to Offset AI Search Damage

• OmniGen 2 blends image and text generation like GPT-4o, but is open source

• Researchers say they may have found a ladder to climb the "data wall"

• Watch: Elon Musk posted on 𝕏 that Grok 4 is planned for ‘'just after July 4"

• Shopify CEO and ex-OpenAI researcher agree that context engineering beats prompt engineering

• Microsoft’s Braga AI chip faces six-month delay, trails Nvidia’s Blackwell

• Hollywood’s pivot to AI video has a prompting problem

• Reddit turns 20, and it’s going big on AI

• Facebook is asking to use Meta AI on photos in your camera roll you haven’t yet shared

• Cloudflare CEO Matthew Prince sees trouble ahead for the open web

• LLM search optimization seems to mirror strategies used in classic SEO, study finds

📄 OctoThinker: Mid-training Incentivizes Reinforcement Learning Scaling

📄 Mind2Web 2: Evaluating Agentic Search with Agent-as-a-Judge ​​

📄 FineWeb2: One Pipeline to Scale Them All -- Adapting Pre-Training Data Processing to Every Language

📄 MADrive: Memory-Augmented Driving Scene Modeling 

📄 LLaVA-Scissor: Token Compression with Semantic Connected Components for Video LLMs

Thank you for reading today’s edition.

Your feedback is valuable. Respond to this email and tell us how you think we could add more value to this newsletter.

Interested in reaching smart readers like you? To become an AI Breakfast sponsor, reply to this email or DM us on 𝕏!