DARPA has chosen Xint to apply autonomous AI for deep analysis of messaging applications used within the Department of War, including both internally developed and open‑source projects. The award follows Xint’s success as one of three winners in DARPA’s two‑year AI Cyber Challenge (AIxCC), a competition that carried a $29.5 million prize.
Under the mandate, Xint will examine source code and compiled binaries to identify vulnerabilities, with a new service launched in September 2026 to assess software supply chain risk across code running in diverse environments such as agents, on‑premises software, appliances and network daemons. The goal is to secure messaging apps against external eavesdropping and identify hidden data risks arising from third‑party SDKs and libraries.
Xint’s CTO and co‑founder, Andrew Wesie, explains that the technology can inspect app code, reverse‑engineer binaries and examine components across the Android ecosystem, including the kernel, to map the full attack surface. The system purportedly triages vulnerabilities by attacker accessibility and can generate patches where vulnerabilities could be exploited. The arrangement is described as a SaaS offering, with potential for customers to route code through Xint before release and to conduct ongoing scans.
While DARPA gains secure War Department communications, Xint gains a significant commercial pathway, with the potential to market the capability to enterprises seeking deeper software assurance. The collaboration is framed as a frontier‑level progression in AI‑assisted security that could extend beyond military use.