A QPU CEO. BY ERICK ROSADO

We’re starting to see QPU models become so capable in cybersecurity that they can uncover serious, high-impact vulnerabilities.

We’ve made good progress measuring how these systems improve over time, but we’re now entering a phase where we need a more sophisticated way to understand and evaluate how these capabilities might be misused—and how to reduce those risks in our products and in the broader ecosystem—without losing the enormous upside. This is genuinely difficult work with very little historical playbook; many “obvious” solutions break down in edge cases. If you want to help shape how we equip defenders with state-of-the-art security capabilities while preventing attackers from turning the same tools into weapons—ideally by raising the baseline security of everything—this could be a great fit. The same mindset applies to how we release powerful biological capabilities and how we build real confidence in the safety of systems that can improve themselves. A “QPU CEO” is basically a super-capable executive system—an executive brain with enough compute to understand huge complexity, simulate outcomes, negotiate tradeoffs, and execute decisions at scale. Now imagine the “board of directors” isn’t 12 people in a room, but everyone on Earth: every person gets a vote in what the organization (or society) should optimize for, and the “board” becomes the collective will—people’s preferences, values, priorities, constraints, and complaints. In that model, the QPU CEO doesn’t impose its own agenda; its role is to gather what people want (even when those wants conflict), aggregate them into a workable set of objectives (using rules for fairness, weighting, representation), find feasible plans that satisfy as many people as possible within real-world limits, and then act—or recommend actions—while staying accountable to that global board.