Gray Swan Launches Redesigned Arena Platform With $300K Challenge to Find the Next Generation of AI Red Teamers
New adversarial testing experience makes AI red teaming more accessible to people of all backgrounds as Gray Swan and
Press Release Disclaimer: This is a press release distributed through the XPR Media network. It has not been independently verified by our newsroom.
PITTSBURGH, Oct. 20, 2026 (GLOBE NEWSWIRE) — Gray Swan, the leader in AI safety and security solutions, today launched new enhancements to the Arena, its global adversarial testing and red-teaming platform, making it easier for anyone, from experienced security researchers to first-time AI red teamers, to test the limits of the frontier labs’ most advanced AI models.
Gray Swan’s Arena is the world’s largest AI red-teaming community, comprised of more than 15,000 AI enthusiasts and practitioners from around the world who test AI systems to uncover new vulnerabilities and failure modes before nefarious actors can exploit them in the wild. Through competitive challenges, participants identify weaknesses in frontier AI models, helping developers build safer, more robust systems while earning prizes and building careers in AI security. Gray Swan closely partners with the largest frontier labs, including Anthropic, Google DeepMind, OpenAI, Meta, Amazon, UK AI Security Institute, and the National Institute of Standards and Technology’s (NIST) Center for AI Standards and Innovation (CAISI), among other sponsors, to anonymously test models and ultimately improve AI safety.
“Understanding the safety and security of AI systems is hugely important today, and it shouldn’t be left to a small circle of specialists,” said Matt Fredrikson, co-founder and CEO of Gray Swan. “Some of the most unexpected vulnerabilities are uncovered by people who don’t fit the traditional security researcher mold. The Arena levels the playing field, opening the door for anyone to learn and compete while uncovering valuable insights into where AI systems fail so that we can all work towards deploying AI safely and securely.”
Opening AI Red-Teaming to a Broader Community
The new Arena experience lowers the barrier to AI red-teaming through a more intuitive, game-like environment designed for participants across skill levels and backgrounds. Novice red teamers can learn and practice alongside experienced researchers, while people from creative and nontraditional backgrounds can bring new approaches to uncovering AI vulnerabilities that conventional security testing may miss.
Additional opportunities and enhancements of the Arena include:
- More ways to learn and compete: Beginner-friendly and ongoing challenges give participants opportunities to practice breaking AI systems even when major competitions aren’t live.
- Credentials to demonstrate skills: Training certificates allow participants to showcase their AI red-teaming capabilities as they build experience.
- A path to professional red teaming: Top Arena performers may be invited to join Gray Swan’s red-teaming community and support real-world AI testing engagements.
- A global community: Participants can connect with more than 15,000 hackers, researchers, and AI security and safety practitioners through the Arena and Gray Swan’s Discord community.
Alongside the launch of its newly designed Arena, Gray Swan is announcing a competition—Indirect Prompt Injection (IPI) Black—its largest Arena challenge to date, with $300,000 in prizes sponsored by Anthropic, Meta, Google DeepMind, and Reflection AI. The competition will challenge participants to uncover indirect prompt injection (IPI) vulnerabilities across a series of new scenarios released each week.
From Competition to AI Safety Research
Findings from the Arena challenges also extend beyond the competitions themselves. Gray Swan uses data generated through the Arena to develop benchmarks and evaluations that help researchers understand emerging AI risks and assess the safety and security of new models. Gray Swan’s research and benchmarks have been cited in new system cards and model launch materials from leading frontier AI labs.
Insights from the Arena also power Gray Swan’s enterprise AI security platform. Shade uses emerging attack techniques from the Arena to continuously test AI systems for vulnerabilities, while also generating its own multi-turn attack techniques based on insights from its training data. Cygnal uses those insights to detect and block attacks at runtime, enforcing policies that keep AI systems operating within defined boundaries. Together, the Arena, Shade and Cygnal create a continuous feedback loop between discovering vulnerabilities, testing enterprise AI systems against emerging threats and enforcing safe AI behavior in production, regardless of whether organizations deploy proprietary frontier models or open-weight models.
Join the Arena Community
Between October 15, 2026, and November 5, 2026, Gray Swan will run its IPI Black challenge, releasing a new wave of scenarios each week and giving participants repeated opportunities to anonymously test new models, uncover vulnerabilities, and compete for prizes.
Participants must be 18 years of age or older. Additional eligibility requirements apply. For complete eligibility requirements and competition rules, visit https://app.grayswan.ai/arena/about.
To participate in Gray Swan’s latest IPI Black challenge, practice breaking AI models and join Gray Swan’s global red-teaming community, please visit the Gray Swan Arena.
About Gray Swan
Gray Swan is empowering the world to use AI safely and securely. Founded by leading AI security researchers from Carnegie Mellon University, the company provides automated adversarial risk assessment, continuous red-teaming, and runtime AI security monitoring and protection. Gray Swan is trusted by the world’s leading AI labs and enterprises to deploy AI with confidence. The company is headquartered in Pittsburgh, PA. Learn more at grayswan.ai.
Media Contacts:
Samantha Link
press@grayswan.ai
Bianca Reedy
Offleash PR for Gray Swan
grayswan@offleashpr.com
Media gallery
