SafeWorld Emerges From Stealth With $12.2M to Build a Safety-Testing Platform for Robots

SafeWorld has emerged from stealth with $12.2 million in seed funding. The startup has built a simulation platform for safety-testing AI-powered robots before they're deployed in the real world. The company is based in Oakland, California.

Founders and vision

Kyle Wong (CEO) and Ding Zhao (co-founder and director of the Safe AI Lab at Carnegie Mellon University) started the company after becoming convinced that AI had changed the game in robotics. "We've been using robots for about 100 years," says Zhao, "but with the arrival of AI, automation has reached a new level. For the first time, robots have been able to step out of their cages and work in the same space as people."

Wong stresses how critical proper testing is: "If you want more capable robots, you need more capable ways to test and validate them. Otherwise you won't deploy those robots."

Funding and market

The funding round was led by Shine Capital and a16z Speedrun. Participants in the round included: • Box Group • Carnegie Mellon University Endowment • Innovation Endeavors • SV Angel • other leading venture capitalists and angel investors

SafeWorld has already attracted customers across several industries: manufacturing, production, logistics, construction, and home applications.

How the platform works

SafeWorld offers a scalable way to test how robots behave around people in rare, dangerous, and unexpected situations without putting anyone at risk. "These are safety scenarios that haven't existed in the world yet," Wong explains. "How do you train a system against something for which there's no large dataset? That's a huge challenge."

The platform lets teams build scenarios directly in the browser based on past incidents, safety standards, and robot logs, with no simulation expertise required. The system runs the robot through thousands of variations with realistic, reactive human movement and measures safety performance.

Because every software update and new environment can introduce new risks, teams rerun tests continuously rather than just once before launch. The result is a complete safety history that engineers, safety specialists, and operational leaders are accountable for.

The platform works across different types of robotics: • robotic arms • humanoids • mobile robots • other types

Wong notes that using simulation for safety requires new approaches: "The difference is in the details and functionality. You need to understand the testing criteria for safety. Different standards exist, but they're incomplete. There are also different types of risk assessment."

When to start working with SafeWorld

The company already works with roughly fifty firms, two-thirds of which are public companies. Wong's recommendation: "If your deployment timeline is less than a year out, start thinking about safety now. If it's more than a year out, it might be too early."

The approach differs between two categories of AI robotics companies. The first: "Bring AI to robots" — companies that take existing, complex robotics hardware and add AI on top. Example: Caterpillar. For them, safety is critical from day one, since the hardware is already operating in the world.

The second: "Bring robots to AI" — startups that start with software. For them, the safety question comes up later, when they move into the real world, but doing the groundwork early optimizes the process.

Thomas Tang, CEO of Anyware Robotics (one of the platform's customers): "Safety has been foundational to Anyware Robotics from day one. Advanced simulation tools like SafeWorld give us a scalable way to test complex scenarios, strengthen our safety processes, and better prepare our robots for real-world deployment."

Plans for growth

SafeWorld is using the funding to: • deepen enterprise adoption • expand the team • advance safety research across the industry • fund and promote research • support open-source projects at the frontier of AI safety

Wong and Zhao agree that third parties help build trust. Wong: "There's the engineering side of safety and the trust side — you need both to deploy. Trust is about stakeholder management and safety approval."

Zhao adds: "This is a deployment problem. Having a third-party company working on this is necessary to earn trust — not just for one company, but for the entire industry."