To prevent artificial intelligence from going out of control, some senior employees at Anthropic are secretly formulating contingency plans. An early employee revealed that there are plans to purchase land and establish hidden locations in remote areas of the United States, so that families can evacuate when AI goes out of control. This idea of 'escaping from AI' is not uncommon within the company. As early as the company's founding period, employees used to simulate hypothetical scenarios similar to the Manhattan Project during gatherings. They also discussed various doomsday scenarios and contingency plans in internal private Slack channels, and adopted the creed that 'everything will never return to peace again.' An Anthropic spokesperson responded that the company currently has over 3,500 employees, with diverse viewpoints and stances on technological discussions within the organization.