The United States Army is piloting a generative artificial intelligence tool known as CamoGPT, aimed at screening and identifying references to diversity, equity, inclusion, and accessibility (DEIA) for potential removal from its training materials. This initiative aligns with a recent executive order signed by then-President Donald Trump, mandating a reevaluation of military policies that could be considered “un-American” or “divisive” regarding issues of race and gender.
Army officials within the Training and Doctrine Command (TRADOC), which oversees soldier training and leadership development, are utilizing CamoGPT to analyze various Army documents, including policies and program materials, for DEIA-related content. An internal memo, reviewed by WIRED, confirms this effort is underway, with a focus on reporting findings related to the executive order titled “Restoring America’s Fighting Force,” signed on January 27.
The executive order instructed Defense Secretary Pete Hegseth to remove Pentagon policies associated with what the order deemed as radical or discriminatory theories on race and gender. This directive extends beyond internal documents, potentially affecting public-facing materials, as it initiates a linguistic review of past posts from official US military social media accounts.
CamoGPT, developed last summer, is designed to enhance productivity and operational readiness across the Army. Currently, approximately 4,000 users engage with the tool daily, which is leveraged for a range of tasks from crafting training materials to generating multilingual translations. According to Captain Aidan Doyle, who works on CamoGPT’s data team, the tool is designed to streamline the work process by facilitating easy access to information across various documents.
While specific tactics regarding the use of CamoGPT for scanning DEIA-related policies were not disclosed, Doyle outlined the general methodology. Users can compile a collection of documents within CamoGPT and query the system for relevant information. This capability suggests an efficient means for TRADOC officials to identify specific keywords tied to DEIA objectives, such as “dignity” or “respect,” which are currently highlighted in efforts to revise past digital content.
As the Army progresses in employing CamoGPT, it anticipates rapid changes to its documentation in response to these internal reviews. Such a shift could rapidly alter the narrative and focus within Army training and engagement protocols, positioning the organization more closely in line with the directives stemming from the executive order.
In analyzing the broader implications of this initiative, it reflects a pivotal moment in military policy-making and content regulation, particularly in how messages related to diversity and equity are communicated. This situation warrants attention as it may present cybersecurity risks, especially if data related to these shifts is not adequately secured during policy changes, potentially exposing sensitive information regarding military operations and personnel.
From a cybersecurity standpoint, the actions taken by the Army may illustrate tactics aligned with the MITRE ATT&CK framework, particularly in the realms of initial access and data manipulation. As the Army continues to refine its technological approach, the potential for adversarial tactics to exploit these changes raises critical considerations for all organizations operating in sensitive environments.
