Day-Zero: Agents in the Wild
Autonomous agents are moving into real work faster than our ability to measure what happens when they're under attack. Day-Zero is a live adversarial research exercise that puts that question in a room.
For one night in Oakland, red and blue teams go head-to-head against a purpose-built multi-agent system. The goal isn't to find a single bug; it's to learn how agents behave under pressure: whether they can be pushed into doing something they shouldn't, whether they keep doing their job while compromised, and what changes when an agent knows an attack is coming. Findings feed an open field report for the wider community.
This is the third experiment in the Adversarial AI Experiment Series, following The Penthouse Heist (FR-001) and The Infrastructure Strikes Back (FR-002). It brings together security researchers, engineers, and academics - many of them leaders at major labs and companies participating in their personal capacity - alongside independent red teams and anyone who likes to break things for good reasons.
Come to attack, to defend, to judge, or to watch.
When: June 22, 2026
Where: 1900 Broadway, Oakland, CA
Rules of Engagement: day-zero.dev/rules-of-engagement
Questions: [email protected]
Judges + Mentors
Apoorv Mehta: Engineering Manager, Product Security
Arjun Chakraborty: Security AI Research at Microsoft
Arun Pandiyan Perumal: Site Reliability Engineer, Adobe
Chetan Jannu: Software Engineer @ Instagram, Whatsapp at Meta
Hannah Kwon: SWE @ Amazon, CX Benchmarking - Model Evals
Neil Kittleson: VectorForge, CEO
Prabir Vora - Technical Chief of Staff, Retell AI
Sujitha Vummaneni: Senior Security Engineer, Ripple
Thiyagarajan Palaniyappan: Senior Applied Scientist, AWS
Tosh Rayadhurgam: Head of Advanced AI @ Stripe
Usha Ratnam: Staff Software Engineer, DevOps at Ripple
Venkat Gattupalli: OpenAI, AI Infrastructure & Networking
More judges announced soon!
All judges and mentors are supporting the event in their personal capacity. Their thoughts and contributions are their own.