Judging at Day-Zero: Agents in the Wild

Judged Day-Zero: Agents in the Wild in Oakland on 22 June 2026, a live adversarial field test presented by Efficient Frontier Labs where teams stress-tested autonomous agents in a shared, multi-tenant environment to see whether they could be compromised, still deliver value once compromised, and whether forewarning offered real defense or a false sense of security. I was among more than twenty judges and mentors, including representatives from Microsoft, Amazon, Adobe, Stripe, eBay, Ripple, and Intuit, evaluating how the agents held up.