In the wake of a hastily brokered deal with the Pentagon, OpenAI has announced a suite of ‘surveillance safeguards’ intended to pacify a public increasingly wary of the military-industrial-AI complex. This pivot, led by CEO Sam Altman, marks a definitive end to the era of AI as a purely civilian tool. As we enter the ‘agentic’ era, where autonomous systems act on behalf of institutions without direct human oversight, the integration of OpenAI’s models into the American military apparatus represents a grave escalation in the automation of violence. The promised safeguards are, in the view of this publication, largely performative. They are designed to provide a veneer of ethical responsibility while the underlying technology is harnessed for ISR (Intelligence, Surveillance, and Reconnaissance) and strategic planning. History shows that when corporate profit motives align with the requirements of the national security state, ‘safeguards’ are the first to be bypassed in the name of operational necessity. The deal with the Pentagon is not merely a business transaction; it is an act of political alignment. By tethering the future of frontier AI to the US military, OpenAI is ensuring that the benefits of agentic systems will be concentrated in the hands of those who hold the most guns, rather than being used to solve the pressing social and environmental crises of our time. The Anti-AI Labour Movement has already begun to flag the dual threat posed by this development: the displacement of intellectual labour and the weaponisation of that same labour against the global working class. We must not be deceived by the language of ‘additional protections.’ In the hands of a hegemon, an autonomous agent is a tool of control, and no amount of silicon shielding can change the fundamental power dynamics of that relationship.