Military drones and robots that rely on artificial intelligence still have no common way to be stress-tested before they reach the field, and a Johnstown, Pennsylvania nonprofit now has Pentagon research money to build one. Concurrent Technologies Corporation (CTC) and its technology transition affiliate, Enterprise Ventures Corporation (EVC), have received a nearly $2 million, two-year contract from the Defense Advanced Research Projects Agency to develop methods for testing and validating AI-enabled autonomous systems in representative mission environments, according to a CTC release dated Oct. 28, 2026. The contract was competitively bid, and it gives CTC its first prime position on a DARPA contract since 2012.
The award is small by defense standards. The problem it targets is not. As autonomy spreads across unmanned aircraft, ground robots and other platforms, the services need confidence that machine-learning models will behave as designed when conditions drift away from the data they were trained on. CTC's release frames the gap directly: "Currently, there is no standardized process to red-team AI/ML systems across diverse mission environments."
What the Nearly $2 Million DARPA Contract Pays For
Under the contract, CTC and EVC will "develop and apply innovative methods to test and validate the performance of autonomous systems in representative mission environments," the company said. The work is meant to assess autonomous systems under realistic operating conditions so that military platforms, which the release identifies as drones and robots, operate safely and as intended across a range of conditions.
Red-teaming, in this context, means deliberately probing an AI system for weaknesses: feeding it inputs, scenarios or environmental conditions designed to expose failure modes before an adversary or the battlefield does. For conventional software, testers can lean on mature verification practices. Machine-learning systems are harder to pin down, because their behavior emerges from training data and can change unpredictably in settings that differ from the lab. A repeatable test method that works across varied mission environments is the piece CTC says is missing.
The release does not name the DARPA program or office funding the work, and GovConFeed is not attributing the contract to any specific DARPA program.
Why CTC's First DARPA Prime Since 2012 Matters
CTC describes itself as an independent, nonprofit, applied scientific research and development professional services organization. Holding a prime position on a DARPA contract again, for the first time since 2012, suggests the organization will work directly with the agency on this effort.
CTC President and CEO Chris Hamilton tied the win to the company's existing AI work. "This award is a testament to the innovative work our team has been doing in AI Assurance and our ability to deliver solutions that address critical national security challenges," Hamilton said in the release. "We are honored to partner with DARPA on this important initiative and proud to contribute to advancing the safety and reliability of autonomous systems for the Department of Defense."
The project builds on CTC research in AI assurance that began in 2018, according to the company, and draws on EVC's production expertise in materials science and engineering. EVC describes its mission as transferring advanced technologies designed and created by others to the marketplace and delivering high-quality, competitively priced products and services to its clients. Pairing a research shop with a transition affiliate suggests CTC wants whatever test methods come out of the work to move toward practical use rather than stay in a report.
How the Red-Teaming Work Connects to DoD AI Policy
CTC positioned the contract as support for the government's push toward safe and responsible AI. The work "supports the U.S. government's commitment to developing safe and responsible AI technologies, as outlined in recent Executive Orders and Department of Defense policies," the release states. It does not cite specific orders or directives.
Dave Saranchak, CTC's executive director for data science, described the goal in terms of predictability. "This project represents a significant step forward in ensuring that AI-enabled systems behave predictably and responsibly in complex mission environments," Saranchak said. "Our team is excited to bring cutting-edge solutions to this effort and to support DARPA's leadership in advancing AI technologies for national defense."
That emphasis on behavior in complex environments matters for program offices. A drone that performs well on a test range may face different lighting, terrain, electronic interference or adversary deception in operation. Testing methods that account for those variables would give acquisition and test communities a stronger basis for fielding decisions.
What Does EVC Bring to the Test Work?
EVC's own website describes the affiliate as providing cost-focused professional services, business consulting and production solutions for government, private industry and the public. It says EVC facilitates the transition of technologies from non-manufacturing entities, SBIR and research-focused organizations, and manufacturers into the industrial base. That is a different skill set from AI assurance research, and the release does not spell out how the two organizations will divide the DARPA work beyond naming CTC's AI Assurance research and EVC's production expertise.
What It Means for Contractors
For firms working in autonomy, test and evaluation, and AI assurance, the award signals that DARPA is still funding foundational work on how to evaluate AI systems, not only on how to build them. Companies developing autonomous platforms for defense customers should expect that testing standards emerging from efforts like this one could eventually shape what program offices ask for in requirements, test plans and acceptance criteria.
The contract also shows that a modest, competitively bid research award can reopen a door. CTC's release says this is its first DARPA prime position since 2012. Smaller research organizations and nonprofits with a track record in a niche such as AI assurance can compete directly for agency work when a solicitation lines up with their specialty.
Teaming opportunities may follow. Red-teaming autonomous systems requires access to platforms, mission data, simulation environments and domain experts. Vendors that build drones, robotic systems, simulation tools or data pipelines could find a role supporting test campaigns, either under this effort or under follow-on work if the methods mature.
Finally, contractors selling AI-enabled systems to the Defense Department should prepare for scrutiny. The release explicitly connects the work to executive orders and DoD policy on responsible AI. Being able to document how a system was stress-tested, and under what conditions it might fail, is likely to become a stronger differentiator in proposals.
