OpenAI, Anthropic researchers quit to write poetry

The exodus of AI safety researchers to literature points to a profound philosophical crisis inside the world's top labs.

• –Resigning researchers argue that scientific frameworks are no longer sufficient to navigate the moral weight of AGI
• –The departures coincide with growing internal anxiety over aggressive commercialization, such as ad testing at OpenAI
• –The shift to the arts highlights a desire for uniquely human connection as AI begins automating intellectual labor
• –Ironically, this cultural phenomenon follows recent findings that "adversarial poetry" can effectively jailbreak leading AI models