Safety? More Like Safety Adjacent: OpenAI's $200K/Year Fellowship to Study What It Already Ditched

OpenAI just dropped a Safety Fellowship offering external researchers $3,850 per week—over $200,000 annualized—plus roughly $15,000 in monthly compute and mentorship, to study what could go wrong with advanced AI. Applications close May 3. Fellows can work from Constellation's Berkeley workspace or remotely. It's like hiring someone to check your parachute while you're actively removing the ripcord, but hey, at least they're paying market rate.

The timing, as they say, is everything.

Just hours before the April 6 fellowship announcement, The New Yorker published Ronan Farrow's investigation revealing OpenAI had dissolved three internal safety organizations over 22 months. The superalignment team was shut down in May 2024 after co-leads Ilya Sutskever and Jan Leike departed. Leike's parting shot: "safety culture and processes have taken a backseat to shiny products." The AGI Readiness team followed in October 2024. The Mission Alignment team was disbanded in February 2026 after just 16 months. For those counting at home, that's three safety orgs going the way of your 2021 NFT floor price—poof, gone.

The New Yorker also reported that when a journalist asked to speak with OpenAI's existential safety researchers, a company representative replied: "What do you mean by existential safety? That's not, like, a thing." Imagine asking your dentist about root canals and getting "what do you mean by teeth" in response. Peak vibes. Oh, and OpenAI quietly removed the word "safely" from its IRS mission statement somewhere in there. Subtle. Very subtle.

To be clear, this fellowship is not a replacement for the teams that walked out the door. Fellows receive API credits and compute resources but no system access—arm's-length research funding, not rebuilding internal capacity. It's like asking a mechanic to write a report on your car while you're holding the keys hostage. Helpful? Maybe. Comprehensive? Unlikely.