DataVeil: Privacy Protection for AI Analysis of Pilot Training Data

Air Education and Training Command’s 19th Air Force Pilot Training Next (PTN) program uses virtual reality, artificial intelligence and data analytics to redefine pilot training. During training exercises, PTN gathers thousands of data points per second from both the simulators and the pilots to create an incredibly detailed profile of each pilot. This rich data is excellent for training and utilizing AI models. However, misappropriation or misuse can easily lead to extremely damaging OPSEC vulnerabilities or violations of airmen’s personal privacy.  PTN leadership has expressed the need for a secure software bridge that connects PTN data (e.g., from learning management systems and learning record stores) to the most advanced commercial AI solutions without compromising OPSEC or data privacy. Any solution would need to be easily interoperable with PTN’s existing Specular Theory training infrastructure, the Apigee data platform PTN is developing, and the many different cloud and on-premise implementations potentially used by different AI vendors. Addressing a data agent interface leads to greater effectiveness for training pilots while driving down cost and supportorting 2030 S&T priorities.  Electric utilities face identical data bridge challenges. Utilities are adopting the Trusted Analytics Chain™ (TAC™) software platform from Via Science, Inc. (VIA) to connect their data to third-party AI/ML applications while ensuring the highest levels of security and privacy.  VIA envisions adding two specific new capabilities to evolve TAC™, which is a TRL-5, into a new application, DataVeil. DataVeil would provide PTN with a secure bridge between sensitive pilot data and commercial AI solutions. In Phase I, VIA will assess the feasibility to adapt TAC™ to (1) improve its AI-enabled data privacy protection related to personally identifiable information (PII), and (2) ingest real-time, microsecond frequency, data. In assessing the feasibility of the proposed DataVeil solution, VIA will collaborate with Massachusetts Institute of Technology (MIT) to maximize the dual-use elements of DataVeil. The conclusion of the VIA and MIT Feasibility Assessment would inform a Phase II project to develop a prototype of DataVeil. This Phase II prototype would provide the foundation for proceeding to Phase III productization and scale with PTN which we have already begun to discuss with PTN leadership. The prototype would also support additional Phase II Add-on and Phase III programs with AF and other DoD departments, while also accelerating the solution’s capture of commercial market opportunities.