Pentagon using Google's TensorFlow APIs to analyze drone footage

Google acknowledges "military use of machine learning naturally raises valid concerns" but says this project "is for non-offensive uses only."
Written by Stephanie Condon, Senior Writer

Video: Pentagon system security flaws can be easily exploited by foreign hackers

The US Defense Department has turned to Google for help analyzing its drone surveillance footage.

Specifically, Google is working on a pilot project with the DoD which leverages TensorFlow APIs to assist in object recognition on unclassified data, Google confirmed to ZDNet.

The partnership is part of the Defense Department's Project Maven, Gizmodo first reported. Launched in April 2017, Project Maven is an effort to use machine learning and AI to analyze the vast amounts of surveillance footage the department captures.

According to Gizmodo, some Google employees last week were "outraged" upon learning about Google's involvement in Project Maven after information about the pilot project was shared on an internal mailing list.

In a statement, a Google spokesperson told ZDNet that this specific project flags images for human review "and is for non-offensive uses only."

The spokesperson added, "Military use of machine learning naturally raises valid concerns. We're actively discussing this important topic internally and with others as we continue to develop policies and safeguards around the development and use of our machine learning technologies."

While Google's involvement in Project Maven is a relatively new development, Google's spokesperson noted that the company has "long worked with government agencies." In 2016, former Google CEO Eric Schmidt (who was at the time chairman of Google parent company Alphabet), was tapped to lead the Pentagon's innovation board, which is tasked with helping the department keep up with current technology.

Editorial standards