AI Military Veterans Secure Patent for Hallucination-Resistant, Explainable AI Technology
A startup founded by retired U.S. military veterans has been granted a patent for technology that improves the explainability and reliability of artificial intelligence. The innovation targets AI hallucinations—false or misleading outputs—and aims to build trust in AI applications across defense, intelligence, and national security sectors.
The company, Data², led by CEO Jon Brewton, focuses on creating AI systems that deliver trustworthy, verifiable results. Brewton emphasizes that organizations in defense and intelligence require near-zero tolerance for error because wrong AI outputs can have serious consequences.
From Military Service to AI Innovation
Data² was founded by Brewton, a former airman, alongside retired Navy SEAL commander Chris Rohrbach and Marine veteran Eric Costantini. The team also includes Jeff Dalgliesh, whose background in oil and gas contributed to the company’s original concept. Their advisory board brings further expertise with figures like former Federal CIO Suzette Kent and Nancy Morgan, former chief data officer of the U.S. intelligence community.
In early 2023, Data² participated in the Catalyst Accelerator program with the Air Force Research Lab and Space Force, launching a proof of concept tailored for the intelligence community. Their military and government experience helped them identify critical gaps in AI technology for mission-critical applications.
Addressing AI Hallucinations Through Data Architecture
Current AI models, despite their advances, often produce hallucinations—outputs not grounded in factual data. Data²’s patented technology tackles this issue by using a “knowledge graph” approach that anchors AI responses directly to verifiable source data. This grounding ensures every AI-generated answer can be traced back to the original data records with citations.
Brewton points out that improving data architecture and structure, rather than solely focusing on better AI models, is key to enhancing explainability and transparency. This approach allows AI systems to be more reliable in high-stakes environments where trust is critical.
Technology-Agnostic and Versatile Deployment
The patented system is designed to work across different AI models and ecosystems. It can be integrated into cloud-hosted environments or air-gapped edge deployments, offering flexibility for various operational settings. This adaptability makes it suitable for a wide range of industries beyond defense, including finance and cybersecurity.
Data² plans to collaborate with major technology providers like Amazon, Microsoft, Dell, and Nvidia to expand its reach. Partnering with these established companies helps bring their explainability and hallucination-resistant technology to government agencies at scale.
Building Trust for Mission-Critical AI Applications
The overarching goal is to enable military commanders and cybersecurity analysts to confidently rely on AI-generated insights in real time. By making AI outputs verifiable and transparent, Data²’s technology supports better decision-making where errors are unacceptable.
This patent marks a significant step toward usable AI in highly regulated and sensitive environments. It underscores the importance of trustworthy AI, especially for governments and organizations that need dependable, explainable solutions.
For professionals interested in expanding their understanding of AI capabilities and applications, exploring courses on explainable AI and AI trustworthiness can provide valuable insights.
Your membership also unlocks:
