Meta is bringing in the founders of Virtue AI, a startup that has carved out a reputation in the AI safety and security space since its founding in 2024. The move signals that even the largest tech companies are willing to poach top talent from specialized startups to shore up their AI defenses.

Virtue AI raised $30 million in funding in April 2025 and counts organizations like Uber and Glean among its clients. Its platform focuses on automated red-teaming, which is essentially stress-testing AI systems to find vulnerabilities before bad actors do, along with compliance tools designed for enterprise customers.

Who are the Virtue AI founders

CEO Bo Li and Chief AI Officer Sanmi Koyejo have been developing adversarial machine learning frameworks since approximately 2014. The team received the NeurIPS Best Paper award and an NSA award for their work on “Decoding Trust,” a framework that evaluates the trustworthiness of large language models. Gartner has also featured Virtue AI in its Hype Cycles for agentic AI and Site Reliability Engineering in 2026.

Virtue AI had positioned itself as a go-to platform for frontier AI labs and major enterprises alike. Losing its founding team to Meta would represent a significant shift in the AI safety talent landscape, concentrating expertise that was previously accessible to a broader set of clients into a single corporate entity.

Meta’s broader AI talent strategy

This hire fits a pattern. In late 2025, Meta recruited Andrew Tulloch from Thinking Machines Lab, another move that signaled Meta’s willingness to pull from specialized AI organizations rather than relying solely on internal development.

Red-teaming has become a critical function at every major AI lab. Virtue AI built an automated version of this process, which scales far better than having humans manually poke at models. Bringing that expertise in-house gives Meta the ability to bake security testing directly into its development pipeline rather than treating it as an afterthought.

What this means for investors

Virtue AI’s $30 million raise in April 2025 is part of a larger trend of venture money flowing into companies that help enterprises deploy AI responsibly. When Meta acqui-hires from this space, it validates the thesis that AI safety isn’t a nice-to-have. It’s infrastructure.

When founding teams leave startups for big tech, the startup’s existing clients may need to find alternative solutions. In Virtue AI’s case, those clients included frontier AI labs and enterprise customers like Uber and Glean, creating openings for competitors in the automated red-teaming space.

The EU’s AI Act, various US executive orders, and emerging frameworks globally all point toward mandatory compliance requirements that will make tools like the ones Virtue AI built not optional but essential.