Gray Swan raises $40M Series A to secure frontier AI
What's the deal? Gray Swan, an AI security startup spun out of Carnegie Mellon University, has raised a $40M Series A co-led by Wing Venture CapitalDealroom has a profile for this one. Try Dealroom → and Madrona. Obvious Ventures, Snowflake VenturesDealroom has a profile for this one. Try Dealroom →, Hudson River TradingDealroom has a profile for this one. Try Dealroom →, Samsung Next, and existing investor Magarac Venture PartnersDealroom has a profile for this one. Try Dealroom → also participated.
The Pittsburgh-based company builds tools that help enterprises and frontier AI labs find and fix unintended behaviour in AI models and agents. It already works with more than 20 customers, including Anthropic, OpenAI, and Meta.
Gray Swan says its benchmarks are cited in 11 recent frontier model system cards — the safety documents labs publish before releasing new models to the public. The funds will go towards scaling the team, go-to-market operations, and deepening partnerships with frontier labs.
Why now? AI agents are moving from demos into production, where they access enterprise data and make autonomous decisions. That shift has massively expanded the attack surface. Prompt injection, jailbreaks, and data exfiltration are no longer theoretical — they're operational realities.
Regulatory pressure is mounting too. As global frameworks tighten, the cost of deploying unsecured AI is rising fast.
Gray Swan's co-founders, Matt Fredrikson and Zico Kolter, spent over a decade at Carnegie Mellon studying how AI systems fail under adversarial conditions. "We have spent our careers conducting research into the safety of AI systems, and we are putting this into practice to ensure that companies can deploy AI with confidence," Kolter said.
What could go wrong? AI security is a crowded and fast-moving space. Gray Swan's edge — deep relationships with frontier labs — depends on those labs continuing to outsource safety evaluations rather than building equivalent capabilities in-house. As models evolve rapidly, keeping benchmarks relevant is a constant race.
The company also faces the challenge of translating research credibility into enterprise revenue at scale. Having 20-plus customers is a start, but the gap between trusted evaluator and indispensable platform is wide.
The signal: Gray Swan's "breakout" growth stage, per Dealroom, underscores how quickly AI security is maturing from niche research into a venture-scale category. The investor mix is telling too — Snowflake Ventures and Hudson River Trading bring enterprise data infrastructure and quantitative trading perspectives respectively, suggesting demand for AI security is expanding well beyond the frontier labs that first adopted it.
Read more: einpresswire.com
