Aptima has completed its Small Business Innovation Research (SBIR) Phase I contract with the US Air Force Research Laboratory, marking a key milestone in the development of OPSHIELD.
OPSHIELD, formed from the Operation Protection System using Holistic Intelligence Extraction from Logistics Data, is a system designed to safeguard operational data.
Go deeper with GlobalData
The system integrates various open-source data streams, employing pattern-of-life analysis to assess whether certain operational information could be vulnerable to enemy exploitation, potentially jeopardising mission integrity.
In current practice, military planners are required to prepare an operations security (OPSEC) annex for every plan they develop.
This includes identifying specific indicators that could potentially be detected by adversaries.
It then requires planners to devise countermeasures aimed at diminishing the chances or impact of any such exploitation.
The company notes that the challenge lies in sifting through the immense volume of publicly available information online, where operationally pertinent data is often just a small subset.
The existing OPSEC procedures are manual and involve scrutinising several distinct datasets that may not individually suggest any movement but collectively could represent a security threat.
During its Phase I efforts, Aptima effectively created and tested large language model (LLM) based tools that allow users to pose OPSEC related queries specific to their organisation’s data environment without necessitating further training or model refinement.
This was made possible by harnessing LLMs’ advanced abilities in interpreting text and extracting annotated relational knowledge.
The OPSHIELD prototype demonstrated proficiency in identifying OPSEC indicators, locating disclosed details, and recommending counteractive measures.
Aptima OPSHIELD project manager Lauren Fortier said: “The OPSHIELD prototype not only demonstrated how large language models can be used to reduce the amount of time currently required by planners to manually parse PAI, but effectively demonstrated zero-shot discovery of OPSEC activity patterns in scenarios it hadn’t previously encountered.”
Aptima’s work enabled this agent prototype to deliver precise responses tailored to unique contexts, even in scenarios not previously analysed by the LLM.
With Phase I funding now complete, there are still avenues for further technological development through subsequent SBIR Phase II and III funding stages.
Aptima seeks additional financial support and collaborative partnerships to propel this innovation towards comprehensive deployment.
Looking forward, Aptima aims to forge alliances with prospective clients, industry peers, and governmental entities to refine and apply this technology within practical settings.
