Another OpenAI exit raises the retention question
Gabriel Petersson's move follows a separate OpenAI hardware-linked departure a day earlier, turning a proof-of-work hiring story into a possible canary moment.
By Ryan Merket · Published
Why it matters
Petersson's exit is small in headcount terms but useful as a signal: elite AI hiring still rewards visible, specific work, even as the labs professionalize around product teams and research roadmaps.
Gabriel Petersson (@gabriel1) has left OpenAI, ending a short run at the AI lab and adding another data point to a widening watch over who is staying inside the industry's most closely tracked AI company.
The timing is what gives the move its weight. Petersson's exit follows a separate OpenAI hardware-linked departure a day earlier, making this feel less like an isolated personnel update and more like a canary-in-the-coal-mine moment for a lab trying to hold technical talent while compressing more of its work around products that can ship.
Petersson's route into OpenAI was unusual even by startup standards. Raised in a small Swedish town, he dropped out of high school at 17 to cofound Depict.ai, an e-commerce data startup, instead of finishing school and pursuing a computer science degree. He later worked at Dataland, a Y Combinator-backed AI startup, before moving deeper into the Bay Area AI scene.
The role that changed his resume was Midjourney. After earlier conventional applications had failed, Petersson built a custom website for Midjourney, recorded a video demo walking through the code, and used the project as evidence that he could do the job. Midjourney hired him as a software engineer in 2023.
That proof-of-work strategy later helped him reach OpenAI, which had previously rejected him. He joined OpenAI in December 2024 as a researcher on the Sora side of the organization, according to the account he has given publicly about his hiring path.
That is why the departure resonates beyond one resume. Petersson had become a small case study in how AI labs could find talent outside the credentialed funnel: build something specific, show the work, make the resume secondary. His exit now sits inside a more uncomfortable question for OpenAI and its rivals: whether the same technical people who fought their way into the labs will stay as the work becomes more organizational, productized, and exposed to public pressure.