Just over two weeks after OpenAI signed a controversial deal allowing the Pentagon to use its AI in classified settings, several concerns remain. While Sam Altman has said the technology cannot be used to build autonomous weapons, the agreement largely relies on the military following its own guidelines. Questions also surround claims that the deal will prevent domestic surveillance.
The company’s motivations are unclear, though factors like rising costs, revenue pressure, and competition with China may be influencing decisions.
How Could OpenAI’s AI Be Used In Military Targeting And Combat Decisions?
As per a report by The Algorithm, despite the agreement being active, OpenAI’s tools are not yet fully integrated into classified systems. Integration with existing platforms, similar to those used by xAI and its Grok model, will take time.
If deployed during ongoing tensions with Iran, the AI could assist analysts in prioritising targets by processing large volumes of text, images, and video.
A defence official suggested that analysts could input potential targets and receive recommendations based on logistics and intelligence data. While a human would review outputs, this raises questions about whether such systems genuinely improve decision-making speed.
The military already uses AI tools like Project Maven to analyse drone footage. However, generative AI could add a conversational layer, offering suggestions on which targets to prioritise. This marks a shift from analysis to influencing real-world actions.
Where Else Could OpenAI’s AI Be Used, From Drone Defence To Administration?
OpenAI has also partnered with Anduril Industries to support counter-drone systems. These tools analyse threats and help intercept attacking drones.
The company maintains that this does not violate its policies, as the systems target machines rather than people. Anduril’s platform, Lattice, already connects multiple defence systems and could integrate OpenAI’s models quickly.
Beyond combat, AI is expanding into administrative work through platforms like GenAI.mil. Tools such as Google Gemini, Grok, and OpenAI’s models are being used for drafting documents and managing logistics.
This reflects a broader push within the Pentagon to embed AI across operations, from battlefield decisions to routine paperwork.