Generative AI is learning to spy for the US military

“We still need to verify the health of the sources,” says Ludon. But unit leaders encouraged the use of large language models, he says, “Because they provide a lot of efficiency during a dynamic situation.”

The Truntic Artificial Intelligence tools that they used by Defeense-Tech Company Vannevar Labs, which in November granted a $ 99 million production contract by the Defense Innovation Unit directed towards the start of the Pentagon to bring intelligence technology to more military units. The company, which was founded in 2019 by the veterans of CIA and the American intelligence community, joins the likes of Palantir, Anduril, and expanding the scope of artificial intelligence as a major beneficiary of the American army’s embrace of artificial intelligence – not only for physical technologies such as drones and self -cars but also to revolutionize how to collect the five -year collection, and data of data clarity.

Although the US military has been developing computers and similar AI’s models, such as the use of Project Maven, since 2017, the use of obstetric artificial intelligence-can participate in a human-like conversation such as those built by Vannevar Labs- represent newer limits.

The company applies the current large language models, including some of Openai and Microsoft, some of which are detailed on its own to Troves from open intelligence source that the company has gathered since 2021. It is difficult to understand the scale that has been collected in 80 different languages ​​in every year in 180. The company says it is able to analyze social media profiles and violate the walls of protection in countries like China to obtain information It is difficult to reach; It also uses non -classified data that is difficult to communicate (collected by human participants on the ground), as well as reports from physical sensors that monitor the radio waves secret to detect illegal shipping activities.

Vannevar then builds the models of Amnesty International to translate information, discover threats, and analyze political feelings, with the results presented through the Chatbot interface that is not different from ChatGPT. The goal is to provide customers with important information on various topics such as international fentanel supply chains and China’s efforts to secure rare Earth minerals in the Philippines.

“Our real focus as a company,” says Scott Phillips, chief technology in Vanifar Labz.

This approach is especially attractive to the American Intelligence Service because for years the world was drowned in more data than analysts who could explain – a problem that contributed to the establishment of Palantir for the year 2003, a company with a market value of more than 200 billion dollars and is known for its strong and controversial tools, including a database that helps in searching for immigrants.

In 2019, Vannevar saw an opportunity to use large language models, which were then new to the scene, as a new solution to data juice. Amnesty International technology can enable not only data collection but actually through an interactive analysis with someone.