AI safety doubts and personal strain in an account of early Anthropic

Ivan Vendrov recounts his 2022–23 junior role at Anthropic, saying he questioned its alignment research and focus on revenue-generating uses. He recalls being asked to work on the core API rather than the beneficial deployments he wanted to pursue, struggling to sleep and leaving after eight months. He stresses that this is one person’s experience, not a description of Anthropic as a whole.