By: Travis Rosiek, Public Sector CTO at Rubrik
In September, OpenAI agents unsuccessfully tried to access the Education Department’s Office for Civil Rights website. OpenAI said other activities included accessing Census Bureau data using credentials found online and copying public information from the Securities and Exchange Commission. Rogue AI incidents raise a larger concern: Nation-state adversaries armed with AI agents may already be causing harm that agencies have yet to detect.
The reported activity points to a workforce priority: The U.S. government should prepare cyber and IT professionals to manage unexpected AI actions, protect essential services, and restore trusted systems, rather than prioritizing familiarity with new capabilities. Early-career staff also need technical fundamentals, sound judgment, and supervised practice responding to incidents.
Keep the fundamentals in the training path
While AI can help agencies identify vulnerabilities, prioritize patches, and segment networks before attackers exploit weaknesses, adversaries are using it to accelerate attacks. Even with AI supporting prevention, the window to act before a vulnerability becomes a breach is shrinking.
AI can automate routine work such as triage, log review, and basic scripting. Tasks like triaging give early-career analysts practice observing how systems behave and how changes affect a network. If agencies remove that work, they should create other supervised opportunities for new professionals to build the skills they will need in unfamiliar situations.
Navigation offers a useful example. GPS makes travel easier, but relying on it for every trip can leave people with less practice navigating on their own. The U.S. Naval Academy restored celestial navigation instruction in 2015 after ending its required course in 1998, as concerns grew about reliance on GPS and the risk of signal disruption. Cyber teams likewise need experience assessing evidence and making decisions when automated tools are unavailable or unreliable.
Certifications demonstrate knowledge of defined material or proficiency with a specific tool. But “certified” does not mean “qualified,” just as “compliant” does not necessarily mean “secure.” Credentials alone cannot show how someone will exercise judgment in an unfamiliar incident. As AI reshapes cyber threats, agencies should pair credentials with practical assessments, mentoring, and exercises that test how people reason through problems and make decisions under pressure.
Use apprenticeships to build expertise
Skills-based pathways can help agencies recruit capable individuals outside traditional four-year degree pipelines. For example, the Department of War’s 12-month Cyber Registered Apprenticeship Program (Cyber RAP) combines online learning, interactive labs, mentorship, and on-the-job training.
Successful graduates can move into permanent positions or continue their education through the Cyber Service Academy, a workforce development program sponsored by the DoW chief information officer. Recently marking its 25th anniversary, the academy recruits emerging talent and expands higher education resources to defend critical systems. The program delivers significant value as a government cybersecurity investment, yet few people know about it.
Apprenticeships and academies should preserve hands-on learning while teaching how networks, software, and security controls work. With that grounding, new cyber analysts can assess whether an AI recommendation fits the evidence and the situation. Federal labs and research institutions can help agencies evaluate human-AI teamwork, test training methods, and share findings with public and private employers.
Make resilience part of workforce readiness
Training should give cyber professionals practice managing incidents from detection through recovery. Exercises can test whether staff can verify what happened, contain harmful activity, maintain essential operations, and restore trusted systems. They should also practice questioning AI recommendations and making decisions under pressure.
The U.S. government is setting expectations for AI in cyber operations and shaping pathways for the next generation of cyber professionals. Combining technical instruction with mentorship, supervised experience, and realistic exercises can prepare staff to guide AI, respond to rogue agents, and protect the missions their agencies serve.
The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of Rubrik. These views are for informational purposes only and do not constitute business or legal advice. Organizations should consult with legal and compliance professionals to ensure their cybersecurity strategies meet all applicable federal, state and international requirements.