OpenAI Distributed Power Rather Than Fill Simo's Role
OpenAI removed Fidji Simo from its chief AGI role in late spring, citing a neuroimmune condition that required medical leave.
The company shifted her into "part-time advisor" status while Kevin Lightcap, who runs the main AGI lab, moved into a parallel "special projects" arrangement. On the surface, this reads as unfortunate timing—a capable leader got sick, and the organization adapted.
But the unstated assumption underneath all of this is that OpenAI has already settled the question of what its AGI safety strategy actually is. If such a settled position existed, replacing the chief AGI officer would be straightforward—you'd promote someone who shared it and consolidate authority.
Instead, the company distributed it. The specific move—making both Simo and Lightcap "special projects" figures rather than filling the chief role—suggests something different. It suggests OpenAI cannot currently agree on that answer. When organizations dodge structural clarity like this, it's rarely because leadership is feeling generous or flexible. It's because the people at the top haven't resolved a genuine disagreement about first principles.
Sometimes distributed authority in a technical domain signals something more honest: an organization is uncomfortable pretending consensus exists when it doesn't.
Simo's transition to advisory status could mean she lost a substantive battle over AGI development priorities, not that she simply fell ill at an awkward moment. Most people read leadership churn as dysfunction. But sometimes distributed authority in a technical domain signals something more honest. An organization is uncomfortable pretending consensus exists when it doesn't. The question isn't whether OpenAI can survive Simo's departure—it's whether the absence of a named successor means the company has actually abandoned the pretense of having one person responsible for deciding how AGI development should work.
And if that's true, who is actually making those calls now?
