Jacob Coxon, who recently departed from Anthropic, has shared his perspective on the current state of AI research and safety concerns. In an interview with WIRED, he discussed the intensive efforts underway at the AI lab and the pressing nature of the work being done there.

Coxon highlighted alignment—the challenge of ensuring AI systems behave as intended—as a critical problem that the industry must address. His comments suggest that the window for making AI systems safe is limited, with labs facing a tight timeline to resolve fundamental safety challenges before systems become more powerful and difficult to control.