Two senior researchers who helped create Google's Veo video generation model have joined Reka, an AI lab focused on building intelligence for robots and physical systems. The merger brings expertise in video generation and multimodal AI to accelerate development of models that understand how the physical world actually works, from motion and physics to temporal dynamics.

Why Does This Matter for AI Development?

The shift represents a meaningful pivot in how AI researchers are thinking about video and simulation. Rather than treating video generation as an end goal, Reka and its new team members are positioning it as a foundation for something more ambitious: building AI systems that can reason about consequences before robots act in the real world. This distinction matters because it moves beyond creating realistic-looking videos to creating AI that genuinely understands physics and motion.

Mateusz Malinowski and Mikołaj Bińkowski, both formerly senior research scientists at Google DeepMind and key contributors to Veo, are now leading research at Reka.

"We're building models that don't just generate video. They understand how the physical world works," said Mateusz Malinowski. "That means simulating motion, physics, and temporal dynamics in ways that enable robots to reason about consequences before they act, enabling a much safer, more responsible AI."

Mateusz Malinowski, Researcher at Reka

The combined team will focus on developing what Reka calls a World Language Action Model (WLAM), an AI system trained on real-world data including egocentric videos and robotics trajectories. Unlike traditional video generators, this model is designed to perceive and act in physical environments by performing realistic simulations for planning.

What Are the Four Pillars of Reka's Strategy?

Reka operates across multiple interconnected areas to support physical AI development:

• Reka Labs: Research and development of foundational intelligence systems designed specifically for the physical world and real-world problem-solving.
• Reka Infer: Inference infrastructure and API tools that allow developers to access multimodal AI capabilities at scale.
• Reka Vision: Video processing infrastructure platform built to tag, reason about, search, and clip large volumes of video data efficiently.
• Reka Claru: Training data specifically curated for physical AI, including egocentric videos, robotics trajectories, world-model footage, and expert human judgment at scale.

This four-part structure reflects a comprehensive approach to building AI that doesn't just generate content but understands and interacts with physical environments.

How Does This Fit Into the Broader AI Safety Conversation?

The timing of this merger aligns with growing industry focus on AI safety, particularly as AI systems become more autonomous and interconnected. On the same day Reka announced the acquisition, Google DeepMind announced a $10 million research funding call focused on multi-agent AI safety, recognizing that as millions of AI agents built by different organizations begin interacting across digital environments, they must do so safely and predictably.

The funding call targets four priority research areas: building sandboxes and testbeds for evaluating multi-agent systems; understanding how collective capabilities emerge in agent networks; strengthening agent infrastructure for secure cross-platform interactions; and developing oversight methods to monitor deployed agent populations. The deadline to apply is August 8, 2026, with awardees expected to be announced in autumn 2026.

Reka's approach to building AI that understands physics and consequences before acting represents one practical answer to these safety concerns. By training models on real-world data and focusing on simulation and reasoning, the company is positioning itself to develop AI systems that can be more predictable and controllable in physical environments.

The merger also brings together researchers from DeepMind, Meta, Amazon, Microsoft, Google, Wayve, and Runway, creating a team with deep expertise across video generation, robotics, and AI infrastructure. Reka is actively hiring across research, engineering, and product roles as it scales development of these physical AI systems.