OpenAI has expanded its policy footprint significantly over the past year, publishing a “blueprint” for AI infrastructure development in November and setting up shop in capitals across Europe.

This morning the company published another such “blueprint,” this time focusing on competition with China and domestic safety concerns. It’s one more example of the biggest AI developers trying to shape the policy conversation around everything from energy to defense and national security, as their technologies inch toward the center of it — and a hawkish second Trump administration is set to take office.

OpenAI has been a major proponent of the idea that AI is a civilizationally game-changing technology (approaching “artificial general intelligence,” in industry parlance). Initially, its CEO Sam Altman loudly stumped in Washington for regulation of the biggest and most powerful systems; more recently the company has been pushing an expansionist line. Now Elon Musk, one of the company’s co-founders (who has his own AI company and has become arguably OpenAI’s most bitter foe), is poised to wield major influence in the second Trump White House, raising questions about what kind of powerful frenemy he might be.