Per John Kiriakou, the U.S. military now purchases equipment and technology using a venture-capital investment model, rather than traditional procurement — a development he raised as evidence that “the writing is on the wall” about the militarization of emerging technology.[1] He frames this as part of a broader shift in which tech and energy corporations have become “military-adjacent,” folded under the protective umbrella of national security.[2]
Kiriakou points to DARPA’s work with Anthropic, OpenAI, and Palantir on technology intended for use “20 years from now” as a further example of this military-venture pipeline.[3] He says the Pentagon’s Department of War recently offered Anthropic a $200 million contract to develop autonomous weapons and mass-surveillance systems, which Anthropic turned down on ethical grounds; OpenAI then accepted the contract after Trump reportedly threatened Anthropic with a hostile takeover for refusing.[4]
Kiriakou has also warned about the risk of powerful private AI models falling into hostile hands. Asked about Mythos, a powerful AI model known for finding corporate vulnerabilities, he said that if a hostile intelligence service such as North Korea’s obtained access to a model like it, the model could be used to hack studios — citing a real-world precedent where a studio’s systems were taken down entirely — or to attack targets such as the Pentagon, the electrical grid, or water purification facilities.[5]