Army launches software security screening platform, with frontier AI to follow

The U.S. Army launched Project Gatekeeper on October 23, 2026, giving commercial developers an automated route to security screening for software intended for Army networks. The Army says the platform provides feedback in near-real time. Access is initially restricted to vendors with military credentials, while frontier AI integration and broader public access remain planned additions.

A shorter evaluation, not an automatic approval

Gatekeeper targets a process the Army describes as especially difficult for small technology companies and vendors outside the traditional defense industry. Previously, getting software evaluated could involve disconnected procedures, existing defense contracts and military credentials. The Army says approval timelines could stretch from six to eight months or longer.

The new platform provides a security sandbox—an isolated testing environment—where developers submit code and containerized applications. Automated tools scan that software, identify vulnerabilities and check compliance requirements. Engineers receive guidance on what to fix, rather than waiting through a manual documentation process before learning which problems need attention.

Those tests run in secure, containerized environments that the Army says prevent unauthorized exposure to operational data. The platform draws its authorization framework from the Army’s existing development, security and operations platform, which provides a commercial-facing connection into the service’s unclassified network ecosystem.

Acting Army CIO Gabe Chiulli says the approach cuts evaluation cycles from months to minutes. That speed claim concerns evaluation, not an automatic authorization to operate, or ATO. Gatekeeper generates standardized compliance evidence for authorizing officials; those officials retain final control over software approvals and acceptance of risk, MeriTalk reported before launch.

Public access is still a later step

The launch does not yet deliver the open commercial entry route described in the Army’s plans. Its current and intended access arrangements are different:

• Available now: Vendors with a military common access card can enter the platform and test their software.
• Planned next: Public access through Login.gov will allow authentication and vetting of people connecting from the commercial internet.

One practical change is the opportunity to test before competing for a contract. DefenseScoop reports that vendors can use Gatekeeper before engaging in an official solicitation, so their code can be vetted ahead of a potential award. That moves security preparation earlier than the point when a company goes on contract with the Army.

Security screening is not the same as finding a buyer. Before launch, Chiulli said vendors must identify an Army customer with an operational need, according to MeriTalk’s October 19 report. Potential sponsors include program offices, system owners and operational units. The CIO’s office would help companies navigate the security pipeline, but would not find customers for them.

Frontier AI will supplement the security tools

The Army plans to add frontier AI models after release to strengthen existing security tools, combine security reporting and help prioritize fixes. Those are supporting functions within the assessment process, rather than a transfer of approval authority. The service has not identified which models it will use, DefenseScoop reported on October 26.

David Wright, chief architect for the Army’s Software and Innovation Center, framed the change at the October 13 AUSA conference as greater access to feedback, not weaker security requirements. His stated aim was to let developers understand what must be corrected before software enters an Army environment. That keeps the rollout’s emphasis on earlier preparation while preserving the security standard.