Collaborative research: cyberai innovation: preparing AI-native software engineers for secure software development

The National Science Foundation Division of Graduate Education awarded Drexel University $249,979 on September 1, 2026, under the STEM Education program (CFDA 47.076) to prepare undergraduate computing students to use artificial intelligence coding tools responsibly and securely in software development.

The project will develop a threat modeling framework for AI-assisted coding workflows and create at least eight hands-on laboratory modules using realistic software development scenarios. Students will inspect agent traces, repository content, tool permissions, and dependency actions to diagnose security problems including prompt injection, unsafe command use, overly broad permissions, credential exposure, dependency confusion, and silent security regressions. The project will assess student learning through pre- and post-assessments and rubric-scored assignments. The resulting educational materials are intended for use by instructors at multiple institutions to prepare students for industry and government needs as AI coding tools become commonplace in software development environments.

Performance takes place in Philadelphia, Pennsylvania, with a period of performance from September 1, 2026, through August 31, 2029. The assistance type is a project grant.

Abstract

This project aims to serve the national interest by preparing undergraduate computing students to use artificial intelligence coding tools responsibly and securely. AI coding tools can now examine software projects, use other software tools, install code packages, and suggest changes with limited human oversight. These abilities can help people build software faster, but they can also create security risks when users do not know how to guide, limit, and check the tools. The project intends to strengthen students? Ability to recognize risky tool behavior, protect software projects, and make sound decisions before AI generated changes are accepted. The significance of this work is that IT will help students build skills needed for safer software development in workplaces that increasingly rely on AI. The project advances education in artificial intelligence and cybersecurity while helping build a stronger technical workforce, drawing on best practices in hands on security education, threat modeling, and evidence based curriculum design. The resulting materials are intended to help instructors at many institutions prepare students for the changing needs of industry, government, and society. The project pursues three connected aims. The first aim plans to define a threat modeling framework for AI assisted coding workflows. The second aim plans to create at least eight hands on lab modules using realistic software development scenarios. In these labs, students are expected to inspect agent traces, repository content, tool permissions, dependency actions, and AI generated code. They are expected to diagnose problems such as prompt injection, unsafe command use, overly broad permissions, credential exposure, dependency confusion, and silent security regressions. The project plans to assess student learning through pre and post assessments, rubric scored assignments, code review tasks, student feedback, and instructor interviews. IT will also study adoption factors, produce a package ready for broad dissemination, and advance understanding of how students transfer security concepts into practical controls for AI assisted software development. This project is supported by the cyberaicorps scholarship for service (cyberai SFS) program which funds proposals to increase national capacity to educate and train professionals in artificial intelligence (AI) or cybersecurity, and to support their placement and retention in the cyberai mission of government organizations. The scholarship track funds academic institutions to award scholarships to students in exchange for their government service. The innovation track seeks transformative education proposals in the areas of AI, cybersecurity, or the integration of AI and cybersecurity. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the foundation's intellectual merit and broader impacts review criteria.- subawards are not planned for this award.