Gray Swan Raises $40M Series A to Secure the AI Systems Everyone Else Is Racing to Deploy

Gray Swan, a Pittsburgh-based AI security company founded by Matt Fredrikson and Zico Kolter, has raised $40M in Series A funding led by Wing Venture Capital and Madrona, with participation from Obvious Ventures, Snowflake Ventures, Hudson River Trading, Samsung Next, and Magarac Venture Partners.

The company builds AI security infrastructure focused on adversarial testing, AI red teaming, model evaluation, and enterprise AI risk management. Its products include Shade, Cygnet, and Arena, a red-teaming network that Gray Swan says includes more than 15,000 researchers working to identify vulnerabilities before attackers do.

The funding arrives as enterprises move beyond AI experimentation and begin deploying AI systems into production environments where reliability, trust, security, and governance become operational requirements rather than research topics.

The broader signal is clear: investors increasingly view AI security as foundational infrastructure. As enterprise AI adoption accelerates, protecting models may become just as valuable as building them.

What Happened

The AI market has spent the past few years obsessed with capability. How smart is the model? How many benchmarks did it beat? How quickly can it reason? The industry has largely behaved like a group of people handed the keys to an exotic sports car who immediately start discussing horsepower while barely acknowledging the brakes.

That backdrop makes Gray Swan's $40M Series A particularly interesting. Founded out of the Carnegie Mellon University research ecosystem in Pittsburgh, Pennsylvania, Gray Swan focuses on a problem that becomes painfully obvious the moment AI moves from demo environments into production systems: intelligent systems can still make dangerous decisions, expose sensitive information, or behave in ways nobody anticipated.

The company's platform centers around three products. Shade provides automated AI evaluation and testing. Cygnet focuses on safety and resilience. Arena serves as a large-scale adversarial testing environment powered by a community of more than 15,000 red teamers attempting to identify vulnerabilities before malicious actors find them first. Wing Venture Capital and Madrona have both built reputations backing infrastructure companies during major technology platform shifts, making their involvement notable beyond the size of the round.

While many AI companies are building intelligence, Gray Swan is building skepticism. And skepticism tends to age remarkably well.

Why This Matters

Every technology cycle eventually encounters reality. The internet encountered cybersecurity. Cloud computing encountered identity management. Mobile computing encountered privacy regulation. Artificial intelligence is now encountering trust.

That transition changes the economics of the market. During early technology waves, investors reward companies that create new possibilities. As markets mature, investors increasingly reward companies that reduce risk. Gray Swan sits directly inside that second phase.

Organizations deploying AI systems increasingly face questions that sound less like software engineering and more like risk management. Can the model be manipulated? Can sensitive data leak? Can an AI agent perform actions it shouldn't? Can safeguards be bypassed? Those questions rarely generate headlines. They generate budget approvals.

That distinction matters because enterprise technology spending often follows fear more consistently than excitement. A company may delay buying new capabilities. It becomes significantly harder to delay addressing security vulnerabilities. This dynamic is one reason frameworks like the NIST AI Risk Management Framework are receiving increasing attention from enterprises building and deploying AI systems.

Market Context

The AI market is experiencing a familiar pattern. Infrastructure is quietly becoming more valuable than attention. The companies receiving the most public discussion are typically building models, applications, and AI agents. Underneath them sits an emerging layer focused on governance, observability, evaluation, compliance, and security.

Gray Swan belongs to that infrastructure layer. The company's Carnegie Mellon roots are particularly noteworthy because academic AI safety research is increasingly finding commercial relevance. Carnegie Mellon has spent decades producing influential research across artificial intelligence, machine learning, and cybersecurity, making Pittsburgh one of the most important but often underappreciated technology ecosystems in the United States.

Ideas that once lived inside research papers are now influencing enterprise procurement decisions. That shift reflects a broader maturation of the AI market. Security is no longer a future problem. Security has become a deployment requirement. Investors appear increasingly willing to fund companies positioned around that reality.

Competitive Landscape

Gray Swan operates within the rapidly growing AI security sector, which includes companies focused on AI governance, model security, adversarial testing, and runtime protection. What differentiates Gray Swan is its emphasis on AI red teaming and community-driven testing.

Arena functions as a continuous feedback loop where vulnerabilities are actively sought rather than passively discovered. The approach mirrors modern cybersecurity practices where organizations assume attacks will occur and prepare accordingly. The question is no longer whether AI systems can fail. The question is whether organizations can identify those failures before customers, regulators, attackers, or journalists do.

That creates growing demand for platforms capable of stress-testing AI systems before mistakes become public incidents. AI red teaming is quickly becoming one of the most important categories within enterprise AI security because organizations increasingly need evidence that models can withstand real-world adversarial behavior.

What This Signals

The most important signal from Gray Swan's funding round isn't the size of the raise. It's category validation. For years, AI safety conversations largely existed inside research institutions, policy circles, and academic communities. Venture capital is now placing meaningful bets on turning those concerns into commercial infrastructure.

That transition mirrors what happened in cybersecurity. The market eventually realized that prevention, monitoring, and resilience could become massive standalone businesses. AI appears to be following a similar path.

The winners won't necessarily be the companies generating the most attention. Some will be the companies quietly ensuring the rest of the ecosystem can operate safely.

The Bigger Industry Shift

Technology markets often begin by celebrating possibility. Eventually they become focused on consequence. Artificial intelligence is entering that phase.

As enterprises move from experimentation to deployment, questions around security, evaluation, governance, trust, and reliability become increasingly difficult to ignore. Gray Swan's $40M Series A reflects investor confidence that those concerns are not temporary friction points. They are becoming permanent components of the AI stack.

The market spent years asking what AI could do. Increasingly, the market is asking what happens when AI goes wrong. Companies that can answer that question may become some of the most important infrastructure providers of the next decade.

Frequently Asked Questions

What is Gray Swan?

Gray Swan is a Pittsburgh-based AI security company focused on adversarial testing, AI evaluation, AI red teaming, and enterprise AI security.

How much funding did Gray Swan raise?

Gray Swan raised $40M in Series A funding led by Wing Venture Capital and Madrona.

Who founded Gray Swan?

Gray Swan was founded by Matt Fredrikson and Zico Kolter, researchers associated with Carnegie Mellon University's AI ecosystem.

What does Gray Swan's Arena platform do?

Arena is Gray Swan's adversarial AI red-teaming network that helps identify vulnerabilities in AI systems through large-scale community-driven testing.

What is AI red teaming?

AI red teaming is the process of intentionally testing AI systems for weaknesses, vulnerabilities, jailbreaks, misuse scenarios, and failure modes before deployment.

Why are investors funding AI security startups?

As enterprise AI adoption grows, organizations need tools that improve security, reliability, governance, compliance, and risk management.

What industry does Gray Swan operate in?

Gray Swan operates at the intersection of AI Security, AI Safety, Enterprise Software, Cybersecurity, and AI Infrastructure.

Why does Carnegie Mellon University's connection matter?

Carnegie Mellon University is widely recognized as one of the world's leading institutions for artificial intelligence, machine learning, and cybersecurity research, making it a significant source of AI innovation and commercialization.