Gray Swan Relaunches Arena 2.0 With $300K Agent Red-Teaming Contest

Gray Swan relaunched its Arena red-teaming platform as Arena 2.0 and opened the Indirect Prompt Injection Black contest, which offers $300,000 in prizes sponsored by Anthropic, Meta, Google DeepMind and Reflection AI. Participants compete by trying to trick AI models into breaking their guardrails, and new attack scenarios will be released each week through Nov. 2. Arena rules require a human to drive each AI agent being tested; anyone caught ignoring an agent that misbehaves is barred from this and future competitions. The platform also offers beginner challenges that run outside major contests, training certificates, and a path for top performers to join Gray Swan's red-teaming community for real-world testing work.

AI-generated from reports · 1 hours agoUpdate

Not enough continuous observation data yet; trend chart is not drawn for now.